A LARGE SCALE PHASED ARRAY ULTRASOUND SYSTEM FOR
NON-INVASIVE SURGERY OF DEEP SEATED TISSUE
by
Douglas R. Daum
Bachelor of Science, Electrical and Computer Engineering
Brigham Young University, 1994
Master of Science, Electrical and Computer Engineering
Brigham Young University, 1995
Submitted to the Harvard-MIT Division of Health Sciences and Technology
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY IN MEDICAL ENGINEERING
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
November 1998
Copyright 0 Massachusetts Institute of Technology, 1998. All rights reserved.
Signature of Author_
Ifarvard-MIT Division of Health Sciences and Technology
November 23, 1998
Certified by
Ai
Certified by
-~--
Kullervo Hynynen, Ph.D.
Associate Professor of Radiology, Harvard Medical School
Thesis Advisor
'-.
H. Frederick Bowman, Ph.D.
Senior Academic Administrator of HST, Lecturer of Radiation Oncology, HMS
Thesis Committee Chair
Accepted by_
Martha Gray, Ph.D.
Co-Director, Harvard- IT Division of Health Sciences and Technology
MASSACHUSETTS INSTITUTE
SCHERING PLOUGH
LIRARIES
A LARGE SCALE PHASED ARRAY ULTRASOUND SYSTEM FOR
NON-INVASIVE SURGERY OF DEEP SEATED TISSUE
by
Douglas R. Daum
Submitted to the Harvard-M.I.T. Division of Health Sciences and Technology on
November 23, 1998, in partial fulfillment of the requirements for the degree of Doctor of
Philosophy in Medical Engineering at the Massachusetts Institute of Technology.
It was demonstrated decades ago that high intensity ultrasound fields can be used
to non-invasively ablate tissue deep beneath the skin without causing damage to overlying
tissue. This was accomplished by using focused ultrasound beams from either a curved
transducer or a flat transducer with an acoustic lens. Unfortunately, the small focal spots
and precise temperature gradients produced using focused high power ultrasound lead to
very small necrosis volumes, making the treatment of large tissue masses (i.e. cancer
tumors) less practical. In addition, until the advent of thermal mapping using magnetic
resonance (MR) imaging, a non-invasive thermal feedback tool was not available to guide
and monitor a thermal ultrasound treatment. This study has investigated the use of
ultrasound phased arrays in conjunction with MR imaging as a method to increase the
coagulation volume of an ultrasound treatment without sacrificing the precision and
control necessary in a clinical environment. The work included the following steps: 1)
the development of phase and power controlled RF hardware to drive large arrays of
continuous wave ultrasound transducers; 2) theoretical simulations and optimizations of
acoustic, temperature, and thermal dose fields to coagulate large volumes of tissue in a
single sonication; 3) extensive testing of array materials and construction techniques
including the design and experimentation of multiple prototype therapy arrays; 4) in vivo
experimental tests of array driving techniques which used a set of temporally multiplexed
acoustic fields in a short period sonication; and 5) in vivo experiments in a large animal
model which demonstrated the feasibility of MR guided ultrasound therapy of liver tissue
using a 256 element phased array. The data indicate that a robust, large scale array can
produce clinically significant volumes of coagulated tissue (5 cm 3 in thigh, 2 cm 3 in liver,
0.5 cm 3 in kidney) in a single 20 second sonication period.
Thesis Committee:
Kullervo Hynynen, Ph.D., Associate Professor of Radiology, Harvard Medical School
and Brigham and Women's Hospital, Thesis Supervisor.
H. Frederick Bowman, Ph.D., Lecturer of Radiation Oncology, Harvard Medical School,
Senior Academic Administrator of Health Sciences and Technology, Massachusetts
Institute of Technology, Committee Chair.
Martin F. Schlecht, Sc.D., Professor of Electrical Engineering, Massachusetts Institute of
Technology, Committee Member.
2
Acknowledgments
I thank my advisor, Kullervo Hynynen, for his tremendous support and guidance on the
preparation of this research.
His insight and contributions have been invaluable both in my
learning and in the overall field of therapeutic ultrasound. I thank him for the confidence and
compassion that he has demonstrated consistently during the research period. He is a role model
that should be emulated. I thank Fred Bowman for all of his help along my doctorate work. Fred
recruited me, guided me to work with Dr. Hynynen, and has served as my mentor as well as my
thesis committee chair. I am appreciative of Marty Schlecht for serving as a member of my
thesis committee and aiding my understanding of quality electrical engineering.
I thank the members of my laboratory that have been trusted colleagues of these years:
Todd Fjield, Mark Buchanan, Nadine Smith, Nathan McDannold, Erin Hutchinson, Pat Lopath,
Xiobing Fan, Sham Sokka, Randy King, Kagayaki Kuroda, Katherine Merrilees, and many
others. Their friendship has made the research very enjoyable.
I am also appreciative of the NIH/NCI (Grant CA46627), the Harvard-MIT Division of
Health Sciences and Technology, and the MIT Department of Electrical and Computer
Engineering for providing funding for this research and for my doctoral education. I thank GE
Medical Systems for the temperature imaging sequence used in this research.
I thank my father and mother who have guided me to strive for my doctorate degree.
Their love can only be matched by my love for them. I dedicate my thesis to my mother who
died of a disease that may someday be treated by techniques developed as part of this research.
Most importantly, I thank my wife, Heather, and my son, Joshua.
They are the most
important people in my life and I appreciate the sacrifices that they have endured so that I could
complete my doctorate degree. They have carried me through the difficult times and shown me
light when there was none seen at the end of the tunnel.
3
TABLE OF CONTENTS
1. IN T RO D U C TIO N .........................................................................................................
11
1.1 Advantages of Non Invasive Surgery ..................................................................
11
1.2 Clinical Example: Liver Tumors..........................................................................
12
1.3 Focused Ultrasound Surgery ...............................................................................
14
1.3.1 D efinition .....................................................................................................
14
1.3 .2 H istory ..............................................................................................................
15
1.3.3
g
............................................................................................
. 16
1.3.4 H yp othesis.....................................................................................................
. 17
1.4 Phased A rrays ......................................................................................................
17
1.4.1 Definition of an Ultrasonic Phased Array ....................................................
17
1.4.2 Ultrasound Phased Arrays for Diagnostics....................................................
18
1.4.3 Ultrasonic Phased Arrays for Therapy ..........................................................
19
1.5 Square Element Spherical Sectioned Phased Arrays...........................................
20
1.6 Optimizing the Treatment of Deep Seated Tissue...............................................
21
1.6.1 Array Geom etry.............................................................................................
21
1.6.2 Transducer characteristics.............................................................................
22
1.6.3 Hardware Requirements...............................................................................
23
1.6.4 Power/Temperature/Dose Considerations for Thermal Surgery.................. 24
1.7 Scope of This Thesis ............................................................................................
25
2. DESIGN AND EVALUATION OF A FEEDBACK BASED PHASED ARRAY
SYSTEM FOR ULTRASOUND SURGERY...................................................................
28
2 .1 Introduction .............................................................................................................
28
2 .2 M ethod s...................................................................................................................
29
2.2.1 Specifications for a Therapeutic Phased Array System.................................
29
2.2.2 Overview of Array Driving System.................................................................
32
2.2.3 System Characterization and Measurement Techniques............... 37
2 .3 Resu lts ...........................--------..................................................................................
38
2.3.1 Class D/E Converter Efficiency ...................................................................
38
2.3.2 Power Output/Regulation of the Class D/E Power Converter ...................... 38
2.3.3 Harmonic Content of Output Sinusoid........................................................
40
2.3.4 Power Measurement Dependence on Transducer Matching......................... 40
2.3.5 Output Phase Response.................................................................................
41
2.3.6 Phase and Power Relationship for a Class D/E Converter........................... 42
2.3.7 Effect of Phase Feedback on Acoustic Fields ...............................................
42
2.3.8p
..............................................................................
. 44
2.4 Discussion...................................................................................................
45
3. A THEORETICAL DESIGN MODEL FOR SONICATING LARGE TISSUE
V OLU M E S.... ........................................................................................................
48
3.1 Introduction............................................................................................
48
3.2 Materials and Methods........................................................................................
49
3.2.1 Area Gain/Axial Attenuation Model............................................................
49
3.2.2 Array Element Design Given the Maximum Focal Volume............. 51
3.2.3 Focal Spacing Simulations............................................................................
52
4
3 .3 Resu lts .....................................................................................................................
3.3.1 Area Gain/Axial Attenuation M odel............................................................
3.3.2 Focal Spacing Analysis .................................................................................
3.3.3 Design Example ............................................................................................
54
54
57
58
3 .4 D iscu ssion ...............................................................................................................
4. ARRAY CONSTRUCTION AND ARRAY MATERIALS......................................
60
63
4.1 Introduction .............................................................................................................
4.1.1 Array Requirements .....................................................................................
4.1.2 Current State of Array Construction ............................................................
63
63
63
4.1.3
p
.....................................................................................
4.2 M ethods and M aterials........................................................................................
65
66
4.2.1 Acoustic Efficiency Measurements...............................................................
66
4.2.2 M aximum Power Measurements......................................................................
67
4.2.3 Inter-element Coupling M easurements ........................................................
4.2.4 Acoustic Field Simulation and M easurement ...............................................
67
68
4.2.5 "Dice-and-Fill" Arrays ................................................................................
68
4.2.6 1-3 Composite Materials..............................................................................
4 .3 Results .....................................................................................................................
71
76
4.3.1 "Dice-and-Fill" Tests...................................................................................
76
4.3.2 1-3 Composite M aterials Tested for Array Construction.............................
4.3.3 Hydrophone Scans of the 256 Element Array ...............................................
4.4 Discussion ..............................................................................................
4.4.1 Non-composite Piezoelectric M aterials for "Dice and Fill" Arrays* ...............
4.4.2 Kerfs Adhesives for "Dice and Fill" Arrays..................................................
4.4.3 Electrical Connections.................................................................................
4.4.4 Composites from M aterial Systems, Inc. .........................................................
4.4.5 Composites from Imasonic...........................................................................
4.4.6 Acoustic Fields from a Large Scale Array .......................................................
4.5 Conclusions.............................................................................................................
5. TEMPORAL SWITCHING TO OPTIMIZE THERMAL DOSE..............................
5.1 Introduction.............................................................................................................
5.2 M ethods and M aterials........................................................................................
5.2.1
5.2.2
5.2.3
5.2.4
Phased Array Design....................................................................................
Acoustic M easurements .................................................................................
Numerical Simulation ....................................................................................
Optimization Routine .....................................................................................
79
87
91
91
92
92
93
95
96
96
98
98
99
99
100
100
101
5.2.5 Switching Rate...............................................................................................
104
5.2.6 Experimental Set Up Using MRI Thermometry ............................................
105
5 .3 Results ...................................................................................................................
10 6
5.3.1 Simulation and W ater Scanned Comparison of Array Fields ........................
106
5.3.2 Optimization Results......................................................................................
5.3.3 M RI Experimental Results.............................................................................
107
112
5 .4 D iscu ssion .......................... .............................
.................................................
1 15
6. A LARGE SCALE PHASED ARRAY SYSTEM FOR MR GUIDED ULTRASOUND
SURGERY IN THE LIVER............................................................................................
5
119
6.1 Introduction ...........................................................................................................
6.2 M aterials and M ethods..........................................................................................
6.2.1 N um erical Simulations...................................................................................
6.2.2 Porcine M odel ................................................................................................
6.2.3 M R Experim ental Set Up ...............................................................................
6.2.4 U ltrasound Surgery Experim ents ...................................................................
6.3 Results...................................................................................................................
6.3.1 In Vivo Thigh Muscle Experim ents................................................................
6.3.2 In Vivo Kidney Experim ents ..........................................................................
6.3.3 Ex Vivo /In Situ Liver Sonication..................................................................
6.3.4 In Vivo Liver Experim ents .............................................................................
6.3.5 H eating Comparison of D ifferent Tissues......................................................
6.4 D iscussion .............................................................................................................
6.4.1 In Vivo Thigh Experim ents ............................................................................
6.4.2 In Vivo K idney Experim ents ..........................................................................
6.4.3 Ex Vivo /In Situ Experim ents ........................................................................
6.4.4 In Vivo Liver Experim ents .............................................................................
6.5 Conclusion.............................................................................................................
7. CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK...............
7.1 Conclusions ...........................................................................................................
7.2 Recom m endations for Future Work......................................................................
8. APPENDIX A: ULTRASOUND DRIVING SYSTEM DOCUMENTATION..........
8.1 Introduction ...........................................................................................................
8.2 System Block Diagram ..........................................................................................
8.3 U ltrasound D riving Cards .....................................................................................
8.4 System Subunits....................................................................................................
8.4.1 Pow er Loop ....................................................................................................
8.4.2 Phasing Loop..................................................................................................
9. APPENDIX B: PARTS LIST AND SCHEMATICS.................................................
9.1 Parts List................................................................................................................
9.2 Schem aticsAppendix C .........................................................................................
9.3 Pressure Calculations for the Phased Array (Zemanek 1971) ...............................
9.4 Intensity and Specific Absorption Rate Calculations (Hynynen 1990).................
9.5 Bioheat Transfer Equation (Pennes 1948) .............................................................
9.6 Therm al D ose Calculation (Sapareto and D ew ey 1984).......................................
10. REFEREN CES..........................................................................................................222
6
119
119
119
120
121
123
124
124
140
142
146
153
154
154
156
157
158
160
162
162
164
167
167
167
169
171
172
197
203
203
205
219
220
221
221
LIST OF FIGURES
Fig. 2-1: Block diagram of the phased array ultrasound driving system
Fig. 2-2: Distributed control architecture of the phased array driving system.
Fig. 2-3: Phase regulation system.
Fig. 2-4: Output power into a 50 W dummy load with and without power feedback.
Fig. 2-5: Acoustic power regulation dependency on proper transducer matching.
Fig. 2-6: Hydrophone scan of acoustic intensity across a focus of the aperiodic array.
Fig. 2-7: Combined array scans with and without phase feedback.
Fig. 3-1: Diagram of the area gain/axial attenuation model.
Fig. 3-2: Maximum spacing in a multiple focus pattern to create a uniform thermal dose.
Fig. 3-3: Results of the area gain/axial attenuation model for a 1.5 MHz array.
Fig. 3-4: Area gain/axial attenuation plots with varying peak sonication temperatures.
Fig. 3-5: Area gain/axial attenuation plots varying frequency.
Fig. 3-6: Thermal dose levels vs. the distance between adjacent foci.
Fig. 3-7: Maximum temperatures simulated for the grid of variable focal spacing.
Fig. 3-8: Simulated contours simulated for a 256 element, 1.1 MHz phased array.
Fig. 4-1: Diagram of a 1-3 Composite piezoelectric.
Fig. 4-2: Planar projection of array elements.
Fig. 4-3: Photographs of the 256 element array.
Fig. 4-4: Electroacoustic efficiency for MS3 material.
Fig. 4-5: Measured and estimate electrical impedances.
Fig. 4-6: Estimated power loss in 8 m cable.
Fig. 4-7: Intensity scans for a single focus shifted off axis 7 mm in the focal plane.
Fig. 4-8: Limits of off axis focal shifting in the focal plane.
Fig. 4-9: Single focus scanned along the axis of the array.
Fig. 4-10: 16 and 25 focus patterns created in the focal plane of the array.
Fig. 4-11: Intensity patterns in the focal plane used for large focal volumes.
Fig. 5-1: Spherical shaped square element array geometry.
Fig. 5-2: Simulated fields for optimization generated using the mode scanning technique.
Fig. 5-3: Switching technique diagram.
Fig. 5-4: Simulated and water scanned fields for 16 element phased array.
Fig. 5-5: Normalized mean square error plots vs. dwell times.
Fig. 5-6: Optimization results of simulated dose across the focal axis.
Fig. 5-7: High perfusion simulation results of static and switched fields.
Fig. 5-8: Homogeneous tissue and inhomogeneous tissue simulations.
Fig. 5-9: In vivo temperature contour images for static and switched sonications.
Fig. 5-10: Lesions from a single four focus pattern and switched focus pattern.
Fig. 5-11: Lesions from a switched pattern lesion and static pattern lesion along axis.
Fig. 5-12: Temperature response in vivo for a static pattern and switched focus pattern.
Fig. 6-1: MR experimental design for porcine experiments.
Fig. 6-2: Experimental on-axis electronic shifting of a single focus.
Fig. 6-3: Lesion produced from the axial shifted sonications of Fig. 6-2.
Fig. 6-4: Temperature images of off axis focusing of a single focus in porcine thigh.
Fig. 6-5: Axial temperature image of a focus shifted off axis.
7
Fig. 6-6: Lesions produced from off axis electrical focusing.
Fig. 6-7: Multiple focus patterns in focal plane of porcine thigh.
Fig. 6-8: Across axis temperature response in thigh muscle for a mid-sized focal pattern.
Fig. 6-9: Images of mid-sized lesion formed in porcine thigh.
Fig. 6-10: Photograph of mid-sized lesion in porcine thigh.
Fig. 6-11: Axial temperature response in thigh muscle for mid-sized focal pattern.
Fig. 6-12: T2 images of lesion formed in sonication from Fig. 6-11.
Fig. 6-13: Temperature images along the array axis for a large focal volume.
Fig. 6-14: End sonication spatial temperature response of the large focal pattern.
Fig. 6-15: Temperature elevations in the focus and in the prefocal tissue.
Fig. 6-16: T2-weighted images of large lesion in thigh.
Fig. 6-17: T2-weighted image of the cross section of three large volume sonications.
Fig. 6-18: Three large focal region sonications close to a muscle interface.
Fig. 6-19: First and second sonications of the ten overlapping sonications.
Fig. 6-20: T2-weighted images of 3.8 x 2.2 x 3.0 cm 3 lesion.
Fig. 6-21: Hematoxylin and eosin stained muscle tissue.
Fig. 6-22: MR images of a kidney sonication.
Fig. 6-23: Photograph of kidney lesion produced from sonication viewed in Fig. 6-22.
Fig. 6-24: Kidney sonication next to vertebrae.
Fig. 6-25: Microscopic slide of kidney glomeruli stained with hematoxylin and eosin.
Fig. 6-26: Large focal region sonication in ex vivo / in situ liver.
Fig. 6-27: MR images of the end sonication temperature and lesion in ex vivo liver.
Fig. 6-28: SPGR image of ribs and temperature image of rib heating.
Fig. 6-29: Rib heating during sonication.
Fig. 6-30: Average temperature elevations in the rib plane.
Fig. 6-31: MR images of a mid-sized focus in vivo liver.
Fig. 6-32: Photograph of liver lesion formed in vivo.
Fig. 6-33: Mid-sized focal sonications at 11, 10, and 9 cm from the array.
Fig. 6-34: MR images of a large focal sonication area in vivo liver.
Fig. 6-35: Photograph of large lesion formed in the liver.
Fig. 6-36: H&E stained tissue of the large thermal lesion of Fig. 6-34.
Fig. 6-37: Liver tissue stained with H&E.
Fig. 6-38: Series of temperature images highly affected by respiratory motion.
Fig. 8-1: Photograph and block diagram of ultrasound driving system.
Fig. 8-2: Photograph of an ultrasound driving system card.
Fig. 8-3: Block diagram for the ultrasound driving system card.
Fig. 8-4: Complete power loop circuitry (from Channel 4 of schematics in Appendix B).
Fig. 8-5: Schematic of the DC-to-RF power converter.
Fig. 8-6: Basic class E amplifier.
Fig. 8-7: FET drain voltage and transformer secondary current.
Fig. 8-8: Superimposed FET drain voltages at 30 W RF output (1.5 MHz).
Fig. 8-9: FET drain voltages at 30 W RF output at 1.1 MHz.
Fig. 8-10: FET drain voltages at 30 W RF output at 1.8 MHz.
Fig. 8-11: FET gate and drain voltages at 30 W RF output at 1.5 MHz.
Fig. 8-12: FET drain and gate at FET turn on time (1W, 1.5 MHz).
8
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
8-13:
8-14:
8-15:
8-16:
8-17:
8-18:
8-19:
8-20:
8-21:
8-22:
8-23:
8-24:
8-25:
8-26:
8-27:
8-28:
8-29:
8-30:
8-31:
8-32:
8-33:
8-34:
8-35:
8-36:
FET drain and gate at FET turn off time.
Voltages at the transformer primary center tap and secondary / filter junction.
DC-to-DC buck converter schematic for Channel 4 on UDSC.
Simplified schematic of a buck converter.
Regulated DC voltage and VSR output voltage.
AC output voltage ripple.
VSR voltage output and filter inductor AC current.
FET drain voltage and AC current.
Schematic of filter for Channel 4 on ultrasound driving card.
Simulated voltage transfer function of filter.
Input and output voltage waveforms for 60 W.
Schematic of the dual directional coupler.
Compensated diode detector circuitry.
Voltage wave forms of output voltage and dual directional voltage.
Input and output of the op amp compensation circuit for a 1 W load.
Block diagram of power feedback loop for transient analysis.
Transfer function equations for transient analysis.
Transient turn on response to 60 W output power using a fixed DC supply.
Power feedback response from a step input change.
Transient response time to 60 W using the power feedback loop.
Phase detection circuitry.
Phase correction circuitry using the 74HCT9046 PLL.
Block diagram for transient analysis of phase correction.
Simulated transient locking response of PLL.
9
LIST OF TABLES
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
2-1: Arrays used to test the ultrasound driving hardware.
4-1: Kerf adhesives tested for diced transducer arrays.
4-2: Prototype therapy arrays.
4-3: Types of coaxial cable used tested in therapeutic arrays.
4-4: 1-3 Piezocomposite obtained from Material Systems, Inc.
4-5: Square element test array results for PZT-4.
4-6: Square element test array results for lithium niobate.
4-7: Interelement coupling dependence on kerf fill.
4-8: Efficiency data for composites from Material Systems, Inc.
4-9: Maximum acoustic power from composites by Material Systems, Inc.
4-10: Inter-element coupling for Material Systems, Inc.
4-11: Electroacoustic efficiencies through 1 m long Belden coaxial cable.
4-12: Electroacoustic efficiencies through 5 m Tensolite coaxial cable.
4-13: Maximum acoustic powers for three transducers using 1 m Belden cable.
4-14: Maximum acoustic powers for three transducers using 5 m Tensolite cable.
4-15: Coupling measurements of adjacent elements.
4-16: Coupling measurements of 6 adjacent elements.
4-17: Electroacoustic efficiency for elements of the 256 element array.
4-18: Electroacoustic efficiency from three elements of the 256 element array.
4-19: Interelement coupling for the 256 element array.
5-1: Phases to create the fields patterns found in Fig. 5-2.
5-2: Optimized switching power levels
5-3: Comparison of switched vs. non-switched fields.
Table 6-1: Relative powers used in the 0.5 x 0.5 cm2 area in the focal plane.
Table
Table
Table
Table
6-2: Relative powers used in the 1.0 x 1.0 cm2 area in the focal plane.
6-3: Summary of large focal lesions.
6-4: Kidney lesions.
6-5:Table of in vivo liver multiple focus sonications and lesions.
10
1. INTRODUCTION
1.1 Advantages of Non Invasive Surgery
Minimally invasive surgeries offer several advantages over traditional surgical
procedures. First, the less invasive techniques lead to faster recovery times and hence
less hospitalization. Second, the reduced use of anesthesia allows the procedures to be
used for a greater number of patients who otherwise could not receive surgical treatment.
Third, infection is less likely to occur as a secondary complication. Lastly, minimally
invasive surgery can reduce overall health care costs. For these reasons, there has been
significant research and clinical application of procedures to minimize a given treatment's
impact on the patient (Gough 1994; Nielson 1995; Passlick et al. 1997; Montori 1998;
Van Natta et al. 1998). This extends to the treatment of benign masses and diseases
which present with either primary or secondary metastases (Holcomb et al. 1995;
Ramshaw 1997).
For example, for qualifying breast cancer patients it is common to
remove a small amount of breast tissue (a lumpectomy) instead of the entire breast (a
mastectomy) (Kinne 1994). Nevertheless, minimally invasive procedures are not without
complication, and some metastatic diseases may be complicated by them (Cirocco et al.
1994; Ramshaw 1997). It is the goal of this research to develop the tools and techniques
for a completely non-invasive treatment of tissue masses that could eliminate even
minimally invasive tumor resection. This can be accomplished through the use of a high
temperature tissue coagulation therapy known as focused ultrasound surgery. To further
11
motivate the use of this procedure, a brief description of a disease whose treatment could
be improved will be presented.
1.2 Clinical Example: Liver Tumors
Each year there are 138,000 patients diagnosed with colorectal cancer in the
United States (Fong et al. 1995). Of this number, about 25% will present at the time of
initial diagnosis with liver metastases, while another 25% will develop liver metastases in
the course of the disease (Fong et al. 1995; Jamison et al. 1997). These metastases are the
direct cause of death for 40,000-55,000 individuals each year (Hughes et al. 1988; Fong
et al. 1995). In addition, although the occurrence of primary liver cancer in the United
States is low compared to the rest of the world, there are about 6,000 cases of
hepatocellular carcinoma diagnosed in the United States each year (Marcos-Alvarez et al.
1996). The untreated patient with either disease has a very poor prognosis and it is rare to
find 5 year survivors (Fong et al. 1995).
The major clinical treatments of primary or secondary liver tumors include
surgical resection, organ transplantation, tumor embolization, chemotherapy, and
radiotherapy (Bruix 1997). Other than organ transplantation, only surgical resection has
offered any proven long term cure (Fong et al. 1995; Marcos-Alvarez et al. 1996; Jamison
et al. 1997; Bruix 1997). Unfortunately, it is estimated that only 10-20% of patients
diagnosed with liver tumors are candidates for resection ((Hughes et al. 1988) indicates
6,000-12,000 patients annually) either due to the extensive spread of the tumors or due to
extra-hepatic disease processes which contraindicate major upper abdominal surgery
(Hughes et al. 1988; Fong et al. 1995). Of those patients that qualify for resection, 20-
12
30% can be cured (Hughes et al. 1988; Fong et al. 1995; Doci et al. 1995; Scheele et al.
1995; Marcos-Alvarez et al. 1996; Jamison et al. 1997).
Nevertheless, many physicians have avoided the use of surgical resection due to
the significant number of complications. Although the percentages vary between clinical
settings, approximately 50% of patients will have complications which will lengthen their
hospital stays beyond the 13 day average (Doci et al. 1995; Blumgart and Fong 1995).
About 35% will have minor complications such as post-operative pneumonia, pleural
effusions, or wound infections while 15% will have more serious complications such as
pulmonary embolism (1%), myocardial infarction (1%), hemorrhage (2-5%), and liver
failure (8%).
The operative mortality ranges from 2-20% depending on the clinical
setting making surgical resection a much less attractive option despite its proven utility
(Fong et al. 1995; Doci et al. 1995; Blumgart and Fong 1995; Marcos-Alvarez et al. 1996;
Taylor et al. 1997).
Due to the low number of liver tumor patients who can undergo resection and the
high number of complications for those who can, several less invasive techniques are
being developed to ablate tumors in vivo.
These include percutaneous injections,
cryotherapy, and ablation using implantable microwave or radio frequency transducers
(Livraghi et al. 1995; McCall et al. 1995; Yamanaka et al. 1995; Sato et al. 1996; Curley
et al. 1997). These techniques have shown promise, but are generally limited to either
small tumors or tumors near the surface of the liver. In addition, all still require an
invasive procedure.
13
1.3 Focused Ultrasound Surgery
1.3.1 Definition
Several researchers have indicated that deep seated tumors in the body can be
treated non-invasively by focused ultrasound applicators (Szent-Gorgyi 1933; Horvath
1944; Burov 1956a; Burov and Adreevskaya 1956b; Oka 1960; Woeber 1965; Clarke and
Hill 1970; Linke et al. 1973; Kishi et al. 1975; Frizzell et al. 1977; Fry and Johnson 1978;
Kremkau 1979; Heimburger 1985; Frizzell 1988; ter Haar et al. 1989; Yang et al. 1991;
Vallancien et al. 1992; Vallancien et al. 1993; Yang et al. 1993; Chapelon et al. 1993;
Sanghvi and Hawes 1994; Prat et al. 1994; ter Haar 1995; Prat et al. 1995; Crum and
Hynynen 1996; Sanghvi et al. 1996; Frizzell et al. 1977). This technique is known as
focused ultrasound surgery (FUS), high intensity focused ultrasound (HIFU), pyrotherapy,
or ultrasound ablation.
Ultrasound is advantageous for two reasons: it can deeply
penetrate soft tissue due to its low absorption rate and it can generate localized
temperature elevations due to its small wavelength. Any target tissue in the body which
is not blocked by strongly reflecting or absorbing materials (i.e. bone or air) can be
accessible to the thermal effects of ultrasound. It is these same characteristics which have
allowed the development of the huge diagnostic ultrasound industry.
Ultrasound can kill tissue through two methods: the temperature elevations caused
by energy absorption or the formation of cavitation bubbles due to high pressures.
Treatments have targeted the use of both of these modalities by operating either below or
above the in vivo pressure threshold of significant cavitation (Hynynen 1991). Although
recent work has investigated the use of cavitation for various treatments (Prat et al. 1994;
14
Hynynen and Jolesz 1998; Sanghvi 1998), the thermal modality is better understood (and
modeled) and tends to offer more predictable lesion sizes and shapes (Hynynen et al.
1996b; Chen et al. 1997; Sanghvi 1998). For that reason, the treatments of this research
have been designed to use the thermal effects of absorbed ultrasound to coagulate tissue.
1.3.2 History
The therapeutic use of high powered ultrasound beams for tissue coagulation was
first accomplished by Lynn in the 1940s (Lynn et al. 1942). In the 1950s, Bill and Frank
Fry at the University of Illinois did groundbreaking work with the application of focused
ultrasound in the treatment of neurological disorders (Fry et al. 1950; Fry et al. 1954; Fry
et al. 1955; Barnard et al. 1956; Fry et al. 1957).
Professor Lele of Massachusetts
General Hospital and then MIT further developed the work in the 1960s and 1970s by
successfully showing that the tissue death was predominantly related to temperature
elevation over a wide range of treatments (Basauri and Lele 1962; Lele 1962; Lele 1967;
Lele and Pierce 1973). Since those initial researchers, others have treated patients with
high power ultrasound in a large number of anatomical locations such as eye (Coleman et
al. 1985), breast (Hynynen et al. 1996; Hynynen 1996a), kidney (Vallancien et al. 1992;
ter Haar et al. 1998b), liver (Vallancien et al. 1992; Wu 1998; ter Haar et al. 1998b),
bladder (Vallancien et al. 1996), and prostate (Vallancien et al. 1992; Gelet et al. 1993;
Madersbacher et al. 1993; Bihrle et al. 1994; Madersbacher et al. 1995; Nakamura et al.
1997; Mulligan et al. 1998; ter Haar et al. 1998b). All of the treatments rely on focused
ultrasound beams with focal volumes typically 1-2 mm in width and up to 10 mm in
length.
Therefore, multiple sonications with appropriate cooling intervals must be
15
implemented to treat large tissue volumes. In addition, there is a decrease of power
deposition control in inhomogeneous material and an appropriate acoustic window
without high impedance media such as bone or low impedance air must be found for the
sonicating transducer. For these reasons, there has been limited success in sonicating
through the abdominal wall due to skin bums and other complications (Vallancien et al.
1992; Yang et al. 1993; Vallancien et al. 1996; ter Haar et al. 1998b).
1.3.3 Image Guidance
Although the therapeutic use of ultrasound was first developed in the 1940s and
1950s, there was no method to non-invasively monitor the treatment or to evaluate its
success.
The lack of treatment feedback has been one of the main reasons that the
therapeutic ultrasound applicators have not progressed as their diagnostic counterparts.
An ideal system could localize the diseased tissue, nondestructively direct the ultrasound
focus to the correct position, monitor temperature during treatment over the entire
temperature range, and verify tissue coagulation following the therapy. Only recently has
the introduction of non-invasive monitoring techniques been developed that show
promise in accomplishing these tasks.
These techniques include ultrasonography (Fry
1968; Fry 1970; Fry 1971; Coleman et al. 1985; Vallancien et al. 1992; Gelet et al. 1993;
Madersbacher et al. 1995; Sanghvi et al. 1996; Seip et al. 1996; Maas-Moreno and
Damianou 1996a; Maas-Moreno et al. 1996b; Simon et al. 1998), computerized
tomography (CT) (Fallone et al. 1982), and magnetic resonance (MR) imaging (Parker
1984; Hall and Talagala 1985; Dickinson et al. 1986; Delannoy et al. 1991; Cline et al.
1993; Darkazanli et al. 1993; Hynynen et al. 1993; Cline et al. 1995; Kuroda et al. 1995;
16
Hynynen et al. 1995; De Poorter 1995; Suzuki et al. 1995; Stepanow et al. 1995; Smith et
al. 1995; Hynynen et al. 1996; Cline et al. 1996; Chung et al. 1996b; McDannold et al.
1998a). Of the three modalities, only MR imaging to monitor FUS has been
experimentally proven in vivo.
1.3.4 Hypothesis
It is the hypothesis of this research that a phased array ultrasound transducer can
ameliorate the disadvantages of single focus transducers by improving power deposition
and increasing the focal volume of necrosed tissue in a single sonication. As one of the
prototype arrays developed in this research was used to demonstrate the feasibility of
using ultrasound phased arrays in a clinical scanner (Hynynen et al. 1996c), MR imaging
was used as a tool to guide the phased array thermal therapy designed in this thesis.
1.4 Phased Arrays
1.4.1 Definition of an Ultrasonic Phased Array
An ultrasonic transducer array consists of a plurality of ultrasonic generators that
have been geometrically configured. These arrays are typically configured in a one or two
dimensional lattice. The vibrating elements can exist in a single plane (known as planar
arrays) or on curved surface (known as non-planar). The size and shape of the individual
elements can also vary between arrays and within arrays. The transducer array can be
constructed by either configuring separate piezoelements or by cutting a single element
into several subelements.
17
Wells divides multiple element diagnostic transducers into two classes (Wells
1977).
The first class is an "array of transducers" that are driven incoherently. The
second class is a "transducer array" that is a group of transducers that are coherently
processed. This coherency regards both the transmission pulse or receiving processing in
a pulse echo system.
In the case of continuous wave transmission, the most basic
ultrasound array can control the amplitude to each ultrasonic generator collectively or
individually. A phased array is defined as an array which can control the vibrational
phase of its ultrasonic elements. It often can control the signal amplitude as well. By
controlling both the amplitude and phase of the generated signals, the shape of the
ultrasound field can be controlled. This is known as beamforming.
1.4.2 Ultrasound Phased Arrays for Diagnostics
The first ultrasonic phased array used in ultrasonic diagnostics appeared in the late
1960s and early 1970s (Somer 1968; von Ramm and Thurstone 1970; Bom et al. 1971;
Thurstone and von Ramm 1974). These arrays were pulse/echo systems which contained
both transmitting and receiving elements. They consisted of a group of delay lines in the
transmission or reception of each element.
Within a decade it was recognized that
apodization (the ability to weight the transmitted signal for individual array elements)
could improve signal response by decreasing undesirable sidelobes (Tancrell et al. 1978;
Eaton et al. 1980; Karrer et al. 1980). Even more improvement in sidelobe reduction
came with the practical implementation of dynamic receive focusing (Reid and Wild
1956; von Ramm and Thurstone 1970; Thurstone and von Ramm 1974). A good review
of diagnostic phased arrays is found in (Thomenius 1996).
18
1.4.3 Ultrasonic Phased Arrays for Therapy
Lynn and the Fry brothers recommended the use of multiple element arrays from
the early days of therapeutic ultrasound (Lynn et al. 1942; Fry and Fry 1960). Multiple
element ultrasound arrays were first clinically implemented for localized hyperthermia (a
thermal technique which raises the temperature of tissue below the point of cell death to
give a synergistic effect to radiation or chemotherapy) at the University of Arizona
(Hynynen et al. 1987).
transducers.
These arrays consisted of overlapping beams of single focus
The beams, however, were not designed to be coherent.
Therapeutic
ultrasound phased arrays were first developed in the 1980s as a method to electronically
steer and focus ultrasound for regional hyperthermia treatments (Do-Huu and Hartemann
1981). The first applicator contained elements formed as concentric rings. It could scan
a single focus along its axis. Cain and Umemura suggested that the concentric ring or a
sector vortex design could also create ring shaped focal patterns to treat the outer
boundary of a tumor (Cain and Umemura 1986). Linear arrays, stacked linear arrays, and
tapered arrays were introduced to scan a single focus in two dimensions (Frizzell et al.
1985; Benkeser et al. 1987; Ocheltree et al. 1987).
Non-planar arrays were then
developed to use the natural focusing of the array geometry combined with electrical
focusing in two or three dimensions.
These included the cylindrical sectioned arrays
(Ebbini et al. 1988), square element spherical section arrays (Ebbini and Cain 199 1a), and
non-planar concentric ring/sector vortex spherical arrays (Fjield et al. 1996; Fjield and
Hynynen 1997). While most of these applicators were initially designed for hyperthermia
treatments, it has been recognized that similar configurations can be effective in high
temperature thermal coagulative surgery.
19
1.5 Square Element Spherical Sectioned Phased Arrays
A particular configuration, the square element spherical sectioned array, has
shown much promise to localize temperature elevations deep within tissue from sites
external to the body (Ebbini and Cain 199 1a; McGough et al. 1992; VanBaren et al. 1994;
McGough et al. 1994; Fan 1995a; Fan and Hynynen 1995b; Wan et al. 1996; Fan and
Hynynen 1996a; Fan and Hynynen 1996b; Hynynen et al. 1996c; Botros et al. 1997;
Daum and Hynynen 1998; Ebbini and Cain 199 1a; Ebbini and Cain 199 1a). These nonplanar arrays are configured on a spherically curved surface. The planar projection of the
array forms a square grid of elements with equal area projections. The configuration was
designed such that a single focus could be electronically steered throughout a three
dimensional volume, multiple focus patterns could be created, and elements on the array
surface that were acoustically blocked from transmitting to the target tissue could be
turned off. Numerical simulations by Fan, et al. (Fan and Hynynen 1995b) showed that a
sixteen element array could decrease treatment time of a large tumor by a factor of four
over a single focus transducer.
To achieve this decrease in treatment time, the
investigator simulated multifocal patterns such that a single sonication induced larger
coagulation volumes.
Methods to generate these multifocal intensity patterns for a
variety of arrays have been presented by (Cain and Umemura 1986; Ibbini et al. 1987;
Ebbini and Cain 1989; McGough et al. 1994). This thesis has built upon the simulation
research on the square element spherical sectioned array to experimentally design,
construct, and test the configuration for application in focused ultrasound surgery for
tumor treatment. This included the creation of a large scale phased array (256 elements)
for the first application of in vivo coagulation of a large animal liver using a phased array.
20
1.6 Optimizing the Treatment of Deep Seated Tissue
Four major categories must be considered to optimize the treatment of deep seated
tissue using a spherical sectioned array:
1.
Array geometry: element size, position, radius of curvature, and aperture
2. Transducer characteristics: frequency and material
3.
Electronic driving signals: generation, resolution, and regulation
4. Power/Temperature/Dose distribution:
pulse duration, total power level, and
power deposition pattern
1.6.1 Array Geometry
The square element spherical sectioned array geometry is similar to the shape of a
single focus, spherical transducer. Indeed the most efficient arrays have been constructed
from single focus transducers. As stated previously, this type of array can steer a single
focus away from the natural focus of the array and can create multiple focus patterns. As
a general rule of thumb, the dimensions of the individual foci are dictated by the array
aperture, the array radius of curvature, and the driving frequency. For a single focus
transducer the well known 3-dB focal intensity length (1) and focal width (w) are given by
F
D
and
l = K 2W
21
where X is the ultrasonic wavelength (about 1 mm at 1.5 MHz in water), F is the focal
distance (radius of curvature) , D is the diameter of the array (aperture), and K, and K2 are
constants which depend on the aperture angle of the array (typically, K, is about 1 for
aperture angles less than 90' and K2 ranges from 8 for a 100 aperture to 14 for a 900
aperture) (ONeil 1949; Wells 1977; Fry 1993). While these equations govern the focal
dimensions of a single focus transducer, they also approximate the dimensions of a
shifted single focus using a phased array as long as the focus is not shifted too far from
the natural focus of the array. On the other hand, the ability to electronically scan a single
focus or create multiple foci is governed by the size, spacing, and number of array
elements. An array is able to shift a focus anywhere within the focal dimensions of the
individual elements (using the same equations as the array geometry with D being the
aperture of the individual element). Therefore, the focal dimensions are set by the array
configuration while the focal shifting and multiple pattern control is set by the
dimensions of the individual array elements. It is noteworthy that the single or multiple
focus patterns are always in the near field of the array but in the far field of the array
elements.
As a part of this thesis, several arrays of the spherical sectioned variety have been
simulated and constructed to investigate the ideal geometry.
These include two 16
element arrays, a 64 element array, a 76 element array, and a 256 element array.
1.6.2 Transducer characteristics
The majority of therapeutic ultrasound arrays have been created from diced PZT
pseudocrystals.
This material has high electrical-to-acoustical efficiency and it is
22
relatively inexpensive. For this reason, the prototype 16, 64, and 76 element arrays were
constructed using this technique by cutting a single focus transducer into individual
elements while taking precautions to safeguard the entire array geometry. This technique
is appropriate for arrays with a relatively small number of elements.
However, to
construct an array with more numerous but smaller elements is more challenging. First, it
is difficult to safeguard the array geometry while dicing a single transducer into many
small pieces since the pieces will shift because they must be air backed for high
transmission (Wells 1977; Hynynen 1990). Second, proper width-to-thickness ratios of
the individual elements must be implemented to avoid undesirable vibrational modes
which decrease the element efficiency (De Silets 1978). Lastly, a robust ground plane is
difficult to create using a diced transducer. For those reasons, this thesis investigated
composite piezoelectrics as an alternative.
piezoelements embedded in a polymer.
dimension.
Composite materials contain small
They can be molded into any shape and
Materials from two different companies were tested for high power
sonications and an appropriate material was chosen for the creation of a 256 element
array. Other issues such as cabling and water sealing were also addressed.
1.6.3 Hardware Requirements
Phased arrays require a great deal more control than their non-phased
counterparts. Circuitry must be able to control both amplitude and phase of the ultrasonic
field independently and with enough resolution to overcome errors due to variation of
transducer impedances, non-uniform phase shifts caused by matching circuitry, and
variable transducer efficiencies.
Previous phased array amplifier systems include the
23
ability to set phase and magnitude of their driving signals, but were unable to accurately
control these parameters for different element sizes and geometry (Buchanan and
Hynynen 1994).
To overcome this inaccuracy, researchers have used hydrophone
calibration routines for the arrays or have implemented invasive hydrophone feedback
(Ebbini and Cain 1991b). Rather than use these techniques, a new type of ultrasound
driving system has been designed as a part of this research. This system uses a DC-to-RF
switching converter (class D/E) to produce RF signals in the 1.2-1.8 MHz bandwidth.
Eight bit power control (0-60 W) for each converter is obtained by varying the DC
voltage from a switching regulator which supplies the power converter.
DC-to-RF
efficiencies of greater than 70% eliminate the need for bulky heat sinks to make the
system more clinically feasible. The power is further regulated in a feedback loop such
that the measured RF power output is used to automatically adjust the DC supply.
RF
output phase shifting of 3600 and better than 2' resolution is achieved by using a
combination of preloadable counters and delay circuitry. In addition, phase feedback has
been implemented to eliminate phase matching errors inherent in arrays using various
sized elements.
As part of this thesis, an amplifier system has been designed,
constructed, and tested in clinical treatments. It can reduce or eliminate the need for
invasive acoustic feedback.
1.6.4 Power/Temperature/Dose Considerations for Thermal Surgery
This research has investigated treatments based on thermal necrosis of tissue
without exceeding the threshold of tissue cavitation.
Previous thermal research was
predominantly based on the steady state bioheat transfer function used in low power
24
hyperthermia treatments. The use of short duration pulses, however, does not allow the
heat within a biological material to reach a steady state due to the long time constants of
perfusion and convection (Davis and Lele 1989; Billard et al. 1990; Hunt et al. 1991;
Dorr and Hynynen 1992). This is a preferable characteristic for thermal surgery since it
makes the treatment less dependent on the perfusion of the tissue. Unfortunately it makes
theoretical calculations more computationally intensive. In addition, the best method to
predict lesion size for a given treatment is by using a non-linear thermal dose empirically
quantified by Sapareto and Dewey (Sapareto and Dewey 1984) (see also (Moritz and
Henriques 1947; Linke et al. 1973; Carstensen et al. 1974; Frizzell et al. 1977)) which in
turn increases computational needs. This research has performed studies to optimize the
dose over a three dimensional field using a prototype array. To optimize the dose, a
temporal multiplexing modality was simulated and experimentally tested.
1.7 Scope of This Thesis
The goal of this research was to develop and test a large scale, spherical sectioned
phased array for the treatment of large tissue volumes in MR guided focused ultrasound
surgery. There were several specific studies that had to be accomplished to reach the
overall goal.
First, a simple theoretical model was developed for evaluating the maximum
necrosis volume from a phased array transducer in a single 10 second sonication. This
model was based on a direct relationship between the output power of the array and a
modeled uniform intensity in the focal region. This simple model offers the ideal case for
power deposition in tissue.
More accurate acoustic models were simulated and
25
application software was developed for an IBM PVS multiprocessor computer and a
desktop PC to optimize acoustic field strength, temperature, and thermal dose in a three
dimensional field using multiple focus fields. This optimization led to the development
of new treatment techniques which used rapid switching between multifocal intensity
patterns to create a "uniform" thermal dose over a given region of interest. In this way
the overall treatment time could be decreased.
Second, the theoretical results were tested experimentally. The construction and
performance of three prototype square element arrays was analyzed. The strengths and
weaknesses of the arrays were illuminated and a determination of the array geometry was
made for the production of a large scale array (256 elements).
Third, to drive the prototype and large scale arrays, new hardware was designed
and constructed. This hardware was designed to drive arrays of varying element shapes.
It contains feedback to ensure both proper power and phase control over the frequency
range of 1.2-1.8 MHz.
The new hardware was implemented with various arrays to
evaluate the response acoustically and the ability to heat tissue in vivo.
Fourth, new array materials were investigated as an alternative to the prototype
PZT-4 transducers. High power efficiency and coupling measurements were compared
between various composite materials until an adequate material was found for the
construction of a large scale array. A 256 element array was then constructed from this
material and the array response was tested acoustically through hydrophone scans and
with acoustic power measurements prior to in vivo experiments.
In conclusion the large scale array was experimentally used in a series of eight in
vivo porcine experiments in the MRI. The array was able to create lesions greater than 5
26
cm3 in the thigh in a single sonication.
Lesions greater than 2 cm3 were generated in a
perfused liver under MR guidance. The results indicate that MR guided phased array
ultrasound surgery is feasible in the liver and that clinically significant volumes of tissue
can be coagulated in short periods of time.
27
2. DESIGN AND EVALUATION OF A FEEDBACK BASED PHASED ARRAY
SYSTEM FOR ULTRASOUND SURGERY
2.1 Introduction
High power ultrasound phased arrays have potential in several therapeutic
applications (Cain and Umemura 1986; Benkeser et al. 1987; Ebbini et al. 1988;
Chapelon et al. 1993; Thomas and Fink 1996; Goss et al. 1996; Hutchinson and Hynynen
These arrays can increase the focal necrosis volume through multiple focus
1996b).
patterns (Ebbini and Cain 1989; Fan and Hynynen 1996a) and electronically steer foci to
reduce the reliance on mechanical positioning systems (Chapelon et al. 1993; Fjield et al.
1996).
The drawback of these arrays is the increased complexity and cost of the driving
hardware.
Although there is a scarcity of published work on the design of ultrasound
phased array driving systems, most designs have used a switching amplifier with duty
cycle control of power (Ebbini and Cain 1991b; Buchanan and Hynynen 1994) and the
use of counters or delay circuitry to adjust the phase (Buchanan and Hynynen 1994;
Lovejoy et al. 1995). These systems have performed well in that they are efficient and
fairly simple. However, new advances in transducer design have reached a point where
the hardware has become a limiting factor for precise field generation. For example,
several array designs have recently been investigated which have elements of different
size and electrical impedance (Fjield et al. 1996; Hutchinson et al. 1996a; Fjield and
Hynynen 1997). Switching amplifiers can not properly drive these arrays without array
specific hydrophone calibration (Ebbini and Cain 1991b) since the output power is load
dependent and the phase of these systems depends on output power level. This chapter
28
will present a system architecture that can accurately drive a therapeutic ultrasound
phased array with various element dimensions such that the need for hydrophone
calibration is reduced. Specifically, the chapter will discuss the importance of distributed
control, electronic element matching, power feedback with a class D/E power converter,
and phase feedback to ensure proper electrical phase at the transducer surface. A more
detailed description of the electronics is found in Appendix A and Appendix B.
2.2 Methods
2.2.1 Specifications for a Therapeutic Phased Array System
2.2.1.1 Therapeutic Transducer Array Description
Phased array transducers of various shapes have been suggested as applicators for
both low and high power therapeutic modes. As described in the introduction chapter of
this thesis, these configurations include annular or concentric ring arrays (Do-Huu and
Hartemann 1981; Cain and Umemura 1986), stacked linear arrays (Ocheltree et al. 1987),
tapered linear arrays (Benkeser et al. 1987), cylindrical sectioned arrays (Ebbini et al.
1988), and square element spherical sectioned arrays (Ebbini and Cain 1991a). While
several of these arrays contain elements which are relatively uniform in size and function,
new arrays such as the aperiodic linear array by Hutchinson (Hutchinson et al. 1996a)
purposely use elements of multiple dimensions to decrease undesirable transmission
grating lobes. The multiple element sizes have an important impact in the design of an
ultrasound driving system-the hardware must be capable of properly controlling the
29
electronic phase and power across transducer loads which have magnitudes in the range
10-10000 Q and varying capacitive phases.
2.2.1.2 Frequency Range
Most therapeutic ultrasound transducers range in frequency from 0.5-10.0 MHz.
The precise frequency and power level is determined by the application. Unfortunately, a
0.5-10.0 MHz frequency bandwidth can only be implemented using less efficient linear
amplifier designs (classes A, B, and AB). These amplifier classes have poor efficiency
and high power dissipation, and therefore require large heat sinks and increased system
weight and bulk. For a large scale array, the system size becomes unreasonable.
By
narrowing the specified bandwidth, more efficient amplifiers can be implemented to
make the system more manageable. In this application, the frequency range was chosen
to be 1.2-1.8 MHz.
2.2.1.3 Power Range
The amount of output power per channel from an array will highly depend on the
size and number of elements in that array. As this system was designed to drive an
arbitrary phased array, it was assumed that the array would have a minimum of 8
elements and a maximum of 1024 elements. The power range was specified to be 0-60
W per channel with 8 bit resolution. The upper power limit, therefore, could be used for
the arrays with a small number of elements while the lower limit would be used for large
scale arrays.
The 8 bit resolution is quadratic with power such that more bits are
available for small power increments.
30
2.2.1.4 Phase Resolution
The amount of phase resolution needed for therapeutic ultrasound arrays is the
topic of some debate. Wang, et al. (Wang et al. 1991) suggest that four bit resolution is
sufficient for ultrasonic phased arrays due to the phase differences presented in
inhomogeneous tissue. Fan and Hynynen (Fan 1995a), however, has shown that higher
resolution is preferable for large scale arrays with complete focal patterns.
For this
reason, the system design will implement a minimum of 8-bit resolution.
2.2.1.5 Control
The control sub-unit of a phased array system performs three essential tasks: it
monitors system performance, it modifies system output, and it sets safety interlocks.
These tasks are interrelated. For example, the system must be able to monitor the output
powers on all of its channels in real time and detect erroneous power levels to ensure
patient safety. If a single array element should fail, the system should be able to quickly
turn off power to that element without disturbing the rest of the array. For arrays with a
small number of elements this can be done with a simple centralized control system.
Monitoring large scale arrays with a single processor, on the other hand, leads to long
communication times and slower response. Similarly, electronic scanning of single or
multiple foci requires that a system be able to rapidly change the output phase and/or
power for all of its channels simultaneously.
A centralized control architecture can
accomplish this for a small array, but the amount of data bandwidth needed to rapidly
communicate with a large number of elements can become unreasonable. For this reason,
this design implements a distributed control architecture.
31
2.2.2 Overview of Array Driving System
A block diagram of the array driving system is found in Fig. 2-1. The system may
be divided into four main units:
1) control and system monitoring, 2) electrical
transducer impedance matching, 3) phase regulation, and 4) power conversion. While
this system is not unlike most phased array systems, the implementation of feedback to
ensure proper phase and power regulation is previously unpublished for therapeutic
ultrasound hardware.
Control System
User Interface
Single Board Computer
Microcontroller
Power Generation
Power Set-Point
Feedback Enable Lin"
DC Power
Supply
Power
Measurement
4-
PhaseRegultionMatched
Shifting
Select
Cneer
Transducer
-- +
Phase
Amplifier Phase Feedback
Detector
Transducer Phase Feedback
Filtering
ci
--
Fig. 2-1: Block diagram of the phased array ultrasound driving system
2.2.2.1 Control Strategy
This system utilizes a distributed control strategy (Fig. 2-2). The basic control
block is the Ultrasonic Driving System Card (UDSC). This 6" x 11" printed circuit board
contains all the hardware to drive four matched transducers. It has a Motorola 68HC 11
32
microcontroller which controls power and phase for the analog hardware, monitors output
signals to trigger safety interlocks, and holds individual calibration data for each channel
on the UDSC.
The cards also contain local read/write memory for a phase and power
stack (more than 250 levels). If it were necessary for the phase and/or power to be
changed rapidly during a sonication, such as if the focus were to be scanned, the phase
and power data can be downloaded directly to the amplifier's local memory prior to the
sonication. A single pulse can then trigger a step in the stack index, and change the
power and phase for the entire array. This dramatically reduces the communication
overhead during sonication, which in turn allows the microcontrollers to more closely
monitor the amplifiers.
Microcontroller
coMemory
Channel
User Interface
External Inputs
Single Board Computer
External Inputs
Microcontrover
o
s memory
1
Channel
->Channel2
Can
3
-Channel
4-
aintroller
-4
io
c
frm
er
1Cane
Channe12Chne2
Channel 3
Channel 3
Cae
Fig. 2-2: Distributed control architecture
4-Chne4
of the phased
array driving system.
All the microcontrollers interface over a single bus with a x486 based single board
computer to report operational status and to receive operational commands from the user
33
interface. This single board computer is dedicated strictly to communicating with the
microcontrollers and interpreting commands from the User Interface. This strategy is
necessary since the User Interface may be occupied with external interfaces such as a
magnetic resonance imager. In addition, the dedicated single board computer allows
more timely communication with a larger number of UDSC (the system is designed to
implement up to 256 UDSC corresponding to 1024 channels).
All of the control
architecture is based on the principle of modularity so that the same hardware may be
used for arrays with different numbers of elements (i.e. individual UDSC may be added
or removed from a given system).
2.2.2.2 Electrical Transducer Impedance Matching
Electrical impedance matching for individual transducers is advantageous for
three reasons. First, matching increases the maximum power transfer from the amplifier
into the transducer. Second, the power delivered into a matched load can be measured
using simple circuitry. Third, matching elements of varying impedance ensures that the
same range of power can be delivered to each individual element in the array. Since
system control and treatment monitoring are essential aspects of this design, the small
increase in circuitry (two passive elements) is justified and easily implemented.
2.2.2.3 Power Conversion
This system implements a class D/E power converter to convert the digital input
signal and a DC source to a high power, high frequency sinusoid. The class D (Baxandall
1959) and class E (Sokal and Sokal 1975) switching amplifiers are based on the same
principle: use active switching devices (FETs) to drive a resonant circuit while avoiding
34
appreciable current flow through the active device when there exists a voltage drop across
it. The theoretical maximum efficiency for each of these converters is 100% (Raab 1977)
although the efficiency decreases as a function of bandwidth and load variation. To
reduce the extraneous harmonic content of the output signal, a low pass filter is added as
an output stage. It is this filter which determines the bandwidth for the system and a
modification of this filter can change the operating range of a given power converter.
Feedback is used to compensate for non-linearities inherent in class D and class E
amplifiers.
The power feedback signal is obtained from a dual directional coupler
(American Radio Relay League 1989; Buchanan and Hynynen 1994) which measures the
forward and reflected power accurately for a 50 Q load. The forward feedback signal is
then fed to a voltage switching regulator which adjusts the DC supply to the class D/E
converter such that the desired RF power is achieved. The feedback signal is also used to
trigger microcontroller interlocks which monitor unreasonably high reflected power (as in
the case of a failed transducer element). More efficient power conversion using a duty
cycle controlled class D amplifier was rejected due to its inherent increase of undesirable
harmonics leading to a decrease in power measurement accuracy.
2.2.2.4 Phase Control
Several methods have been proposed to phase shift the output signal (Houghton
and Brennan 1992; Cook 1993; Lovejoy et al. 1995). The simplest method uses preloadable counters similar to those used by Ngo (Ngo 1988). Unfortunately, as pointed out
in Lovejoy, et al. (Lovejoy et al. 1995), 8-bit phase resolution using this technique
requires a master clock frequency 256 times the ultrasound frequency, increasing
35
complexity and decreasing reliability. Lovejoy, et al. (Lovejoy et al. 1995), therefore,
recommends the use of a discrete delay based system.
This system implements a
combination of both counters and delay circuitry. Fig. 2-3 is a diagram of the phase
regulation unit.
The input master clock operates at 16 times the frequency of the
transducer (e.g. 24 MHz for a 1.5 MHz transducer). This clock is applied to simple preloadable four bit counter to create phase steps of 22.5 degrees. The other four bits of
resolution are created using a delay chip (8 bits of 0.5 ns steps). This combination of
counters and delay circuitry is effective because it increases phase resolution while
avoiding ultra high frequency master clock signals and a significant increase in chip
count.
Input Phase to
Digital
OperatidengyPhase
Frequency x 16
Divide by
S 16
Counters
Delay
Chip
Set
---
Point
->
PaePower
Lock
Loop
Converter
Output Phase
from Power
Converter or
Transducer Face
Fig. 2-3: Phase regulation system.
Like power control, feedback is necessary to ensure proper phasing of a class D/E
amplifier. This method uses a phase locked loop (PLL) based feedback loop to adjust the
input digital clock of the power stage to regulate the phase of the high power output sine
wave (Sowlati et al. 1995; Sowlati et al. 1996). The feedback signal can be obtained
from either the matching circuitry of the transducer or directly from the transducer face so
that the matching delay is eliminated.
36
2.2.3 System Characterization and Measurement Techniques
2.2.3.1 Measurements into a 50 Q Load
A Bird 50 Q, 200 W dummy load was used to characterize the system. Individual
channel efficiencies were calculated as the RF power delivered to the load divided by the
DC power to the system. In all cases, the RF power was measured using a Hewlett
Packard 438A Power Meter with a Werlatone (C1373) coupler. The system frequency
response was measured using an Hewlett Packard 8590A Spectrum Analyzer and
waveform measurements were recorded using a Tektronix TDS 380 Oscilloscope.
Transducer impedances were measured using an Hewlett Packard 4193A Vector
Impedance Meter.
2.2.3.2 Measurement in transducer loads
The ultrasound driving system was experimentally tested using several transducer
arrays (see Table 2-1 for descriptions of arrays). The arrays contained between 14 and 62
elements with multiple element sizes in each array. The variety of transducer elements
was used to demonstrate the capability of the system to control power and phase with
several element sizes and shapes. The unmatched transducer impedance values ranged
between 20 and 1000 Q in magnitude and were always capacitive.
Acoustic
measurements were made with a 0.5 mm hydrophone (Precision Acoustics, LTD) or with
a radiation force technique (Stewart 1982).
37
Table 2-1: Arrays used to test the ultrasound driving hardware.
Array Design
Number of
Frequency
Reference
1.64
1.5
1.5
1.07
(Daum and Hynynen 1996)
(Fjield and Hynynen 1997)
(Fjield et al. 1996)
(Hutchinson and Hynynen
1996b)
Elements
Spherical Sectioned
Sector/Concentric
Concentric Ring
Aperiodic
16
52
14
62
2.3 Results
2.3.1 Class D/E Converter Efficiency
At 1.5 MHz the DC-to-RF efficiency was measured to be 78% at 60 W, and
dropped to 68% for output levels below 2 W. The main losses occur in the voltage
switching regulator and the ferromagnetics of the filter of the class D/E converter.
To
further improve efficiency, larger magnetics could be used, but the added bulk of the
magnetics would be greater than the decrease in required heat sinking.
Similar
efficiencies are found throughout the operating bandwidth.
2.3.2 Power Output/Regulation of the Class D/E Power Converter
The effect of power regulation is illustrated in Fig. 2-4. For a desired output of 10
W into a dummy load, the power regulation feedback lowers the maximum error from
20% to 1% in the specified amplifier bandwidth (1.2-1.8 MHz). For frequencies below
the system bandwidth (1.2 MHz) the regulation yields larger errors due to the higher
harmonic content of the output signal.
The high frequency limit of the amplifier is
determined by the maximum specified output power.
This power level drops rapidly
when operating above 1.8 MHz due to the low pass filter cutoff.
38
14
12
10
o
6
-.-
No Power Feedback
EIW
--
0
.
Power Feedback
2
1 .2
1.4
1.6
1.8
Frequency (MHz)
Fig. 2-4: Output power into a 50 Q dummy load with and without power feedback. When feedback
was not used, the DC supply to the power converter was set such that at 1.5 MHz the output power
would be 10 W.
Although transducers are matched to 50 Q, a tuned amplifier such as class D or
class E will still suffer a variation in power due to different transducer impedances off
resonance. For example, each element of a 16 square element array (Daum and Hynynen
1996) was matched to 50 Q at the array's resonant frequency (1.64 MHz). When each of
its elements was driven individually with the same amplifier with a fixed supply voltage
(no power feedback), the measured output power varied 20% (4.75-5.80 W).
By
implementing power feedback, the output power variation decreased to less than 1%
(5±0.08 W).
39
2.3.3 Harmonic Content of Output Sinusoid
In the frequency band 1.2-1.8 MHz, the highest harmonic measured while driving
a 50 Q dummy load is 36 dB lower than the primary signal (this occurs at 1.2 MHz).
When the harmonics are greater than -30 dB (at frequencies below 1.2 MHz) the power
measurement capability of the system is decreased and the power regulation has
decreased efficacy.
2.3.4 Power Measurement Dependence on Transducer Matching
To correctly measure and regulate power using a dual directional coupler
(American Radio Relay League 1989), the transducer must be matched using LC circuitry
such that its impedance at the operating frequency is 50 Q (the standard impedance of a
dual directional coupler). The coupler yields two output signals representing the forward
and reflected power delivered to the load. If the impedance is exactly 50 Q there is no
reflected signal and the forward power accurately measures the power delivered to the
load. If the impedance varies from 50 Q, the measured forward power will be greater
than the actual power delivered to the transducer. By regulating the output power using
the measured forward power, the system will never deliver more than the specified
power. To test this, the matching circuit of a transducer was varied such that the load
impedance differed from the ideal 50 Q. The acoustic output power of the mis-matched
transducer was then measured using radiation force measurements for a constant
regulated power level.
The results are plotted in Fig. 2-5 with the center contour
indicating the theoretical point where the actual acoustic output power is 10% less than
the desired power (see Appendix A for theoretical details). Power regulation, therefore,
40
guarantees that the power delivered to a mismatched transducer will not exceed the
programmed output power. This is important because it avoids the chance of sonicating
excessive powers in a clinical treatment in the case of a broken transducer.
200
L: -73%
180
_
4
160K: -45%
140 -
H: -15%
120 -
J: -20%
100
80
o
60-
B: -5%
A:
D:-21%
40
0%
C: -19%
20F: -69%
-80
-60
__________
-40
E: -2%
,
0
20
-20
G: -74%
40
60
80
Load Phase (Degrees)
Fig. 2-5: Acoustic power regulation dependency on proper transducer matching. The contour lines
mark theoretical 10% increments where the actual power output is lower (by that percentage) than
the set point power. The letters plot the experimentally mismatched transducer impedances whose
output power was measured for a given set point power. The percentage indicates the drop in
acoustic power vs. the properly matched 50 Q transducer load. There is good correlation between
theoretical and experimental results.
2.3.5 Output Phase Response
The output phase is characterized by three parameters: range, resolution, and jitter.
The output range of the system is 360' for all frequencies in the bandwidth. The phase
resolution is 0.5 ns (0.270 at 1.5 MHz) resulting from the delay circuitry. The output
phase, however, has some jitter caused by the locking of the PLL. This jitter ranges
41
3-8 ns across the frequency bandwidth resulting in an uncertainty of 2-3' in the phase of
the signal.
2.3.6 Phase and Power Relationship for a Class D/E Converter
A class D/E power converter does not maintain the phase of the input over the
entire output range of the amplifier (Sowlati et al. 1995). This means that the output
phase will depend on the output power level.
As a typical class D or class E amplifier,
the phase varies 480 from 0 to 60 W in this system without feedback. Feedback from
either the amplifier output or the transducer face reduces this error to less than 3'.
2.3.7 Effect of Phase Feedback on Acoustic Fields
To achieve maximum power transfer and to accurately measure output power in
this system, transducer loads must be matched to 50 n. Unfortunately, the matching
network introduces a phase shift between the amplifier output voltage and the transducer
(Ebbini and Cain 1991b). If all array elements are exactly the same impedance then this
shift is constant and unremarkable. If the elements are different shapes or sizes, this shift
will vary (Ngo et al. 1989; Ebbini and Cain 1991b). For example, the measured phase
shift ranged from 29-94' for a concentric ring array (Fjield et al. 1996) and 300 for an
aperiodic array (Hutchinson and Hynynen 1996b). Phase feedback using the transducer
voltage as the feedback signal automatically compensates for these shifts. Fig. 2-6 and
Fig. 2-7 illustrate the acoustic effects of applying phase feedback. A line scan across the
single focus of the aperiodic array (Fig. 2-6) demonstrates that phase feedback from the
transducer face for a simple linear array can increase peak focal intensities by 25%
42
compared to no feedback and 18% for feedback that is taken from the amplifier output
Fig. 2-7 contains a contour plot of the acoustic fields measured using a
stage.
hydrophone for the combined sector vortex/concentric ring array. The implementation of
phase feedback decreases undesirable foci and increases the desired peak intensities by an
average of 20% indicating an improvement in the control of the acoustic field.
1 -
Phase Feedback
:.Transducer
0.9
Cl)
-Amplifier
:.
0.8 ---
08
Output Phase Feedback
No Phase Feedback
o 0.7
0.6 -0
O
V
N
E
0
0.50.4 - -
0.3 -0.2 -
-10
-8
-6
-4
-2
2
0
4
6
8
10
Distance (mm)
Fig. 2-6: Hydrophone scans of acoustic intensity across a single focus for the aperiodic array. The
focus was located 4 cm from the center of the array and the scan was performed parallel to the array
at that depth. The scans were repeated for each type of phase feedback at the same power and phase
inputs. Intensities were normalized using the peak measurement of the three scans.
43
4-
E
E
-2
0
0
-4-
-4
2
-2
4
X-Coordinate (mm)
4-
E
2-_
-4-
-4
-
-2
X-Coordinate (mm)
Fig. 2-7: Multiple focus hydrophone scans across the focus of a concentric ring/sector vortex array
without phase feedback and with phase feedback. The acoustic intensity peaks of the feedback scan
are 20% higher on average than those without feedback. The contour lines correspond to equivalent
acoustic intensity amplitudes.
2.3.8 System Response Times
As stated previously, this system has memory for each microcontroller (see Fig. 22) which can be preloaded with a stack of power and phase settings. A single pulse on
the bus triggers a change in output phase and power set by the values of that stack.
44
Following that trigger, output power will settle within 1% of a step input of 1 W and 10
W in 175 ps and 22 ms respectively. The power feedback is almost critically damped so
there is minimal power overshoot. The output phase with feedback locks within 250 ps.
Therefore an accurate system output occurs in less than 250 pLs for a 1 W step input and
22 ms for a 10 W step input. The next stack value is available from the microcontrollers
within 20 ms such that another trigger pulse may be received.
If a faster response time is needed then the system can operate with a disabled
power feedback loop. The power settling time is then 5.6 pts and 9 pts for a 1 W and 10
W step input. This makes the system response time approximately 250 pIs for all output
power levels with 20 ms needed for the microcontrollers to update the stack (250 stack
levels).
2.4 Discussion
The ultrasound array driving system described in this paper is able to accurately
produce RF signals of appropriate power and phase for arrays of multiple element sizes
and frequencies without requiring array specific calibration.
This system marks an
improvement both in ultrasonic control and in patient safety. By implementing phase and
power feedback, the nonlinearities of previous systems can be alleviated without
necessitating a change to a less efficient amplifier design. Due to their high efficiency, a
256 channel system is about the size of a filing cabinet, making this system suitable for a
clinical setting. The distributed control architecture gives the system a fast response time,
allowing for proper treatment monitoring and electronic focal scanning for a large scale
array.
45
An important aspect of this design is the measurement of the power delivered to
each transducer element.
Since tissue necrosis is a logarithmic function of temperature
(Sapareto and Dewey 1984), small errors in power generation and/or measurement can
greatly affect the tissue response.
For that reason, individual power measurement is
needed for each array element to accurately control treatment conditions.
Simply
measuring the total output power for entire arrays cannot offer this critical information.
As a result of improved and individual power measurement, automatic power
regulation is now possible. It has been shown that the variance in output power between
different transducers driven by class D or E amplifiers can be decreased by implementing
a simple feedback loop. This is especially important for phased array fields which rely
heavily on destructive interference of same magnitude fields (Cain and Umemura 1986;
McGough et al. 1994). By measuring the power into a matched 50 Q transducer load, the
variation of acoustic power output is limited to the variation of the electroacoustic
efficiency of the elements-a characteristic which is easily measured with the radiation
force technique. Power regulation also acts as a safety feature since the output power of a
mismatched transducer will be regulated at or below the desired output power level.
Similarly, phase feedback is important to the operation of arrays whose elements
have varying sizes or output power requirements.
Without feedback, the user loses
control of accurate phasing for a class D or E amplifier between multiple power levels
and hence the ability to precisely control the acoustic field patterns.
An uncalibrated
variable delay caused by the matching circuitry for transducers can also reduce array
performance.
Both of these sources of error are overcome by the implementation of
simple feedback circuitry. Therefore, if an array's elements are properly aligned and the
46
time delay between the transducer's electrical field and mechanical vibration is uniform,
then this technique can eliminate the need for amplitude and phase correction with a
hydrophone.
This technique, however, does not eliminate phase errors caused by
differences in the acoustic properties of the transmission medium and in some cases a
hydrophone correction technique may still be useful (Ebbini and Cain 1991b).
Nevertheless, for the arrays tested as part of this research, acquiring a phase feedback
signal from the transducer face can directly improve acoustic intensities by 20-25%
without any array calibration or hydrophone feedback.
47
3. A THEORETICAL DESIGN MODEL FOR SONICATING LARGE TISSUE
VOLUMES
3.1 Introduction
Focused ultrasound surgery depends on a high intensity gain from the transducer
surface to the focal volume to cause significant temperature rises only in the target tissue
volume. A single focus therapeutic transducer can typically generate temperatures at their
focus high enough to coagulate tissue in seconds with an effective thermal focus volume
of 1-2 mm in width and 5-15 mm in length. As stated earlier, this makes the treatment of
large tumor volumes extremely time intensive since a delay between successive
sonications is needed to avoid near field heating (Damianou and Hynynen 1993).
Spherically shaped ultrasound phased array applicators offer a modality to lessen this
extensive treatment time for larger tissue volumes by using multiple focus patterns to
create an effective focal size much larger than an individual acoustic focus (Ebbini and
Cain 1991a; Chapelon et al. 1993; Fan and Hynynen 1995b).
This chapter presents a
study which theoretically explored the limits of a spherically shaped array to coagulate
large tissue volumes in a single sonication and used a simple model to predict the
maximum necrosis volume for any array geometry given the target tissue depth and
available acoustic window. From this data the array designer can determine the minimum
number and maximum size of array elements required to generate the desired effective
48
focal volume for thermal surgery. A theoretical analysis of the focal spacing to create a
uniform thermal dose distribution using multiple focus fields was also performed.
3.2 Materials and Methods
3.2.1 Area Gain/Axial Attenuation Model
The area gain/axial attenuation model is diagrammed in Fig. 3-1.
This model
estimates the focal power deposition intensity (Pf) at any distance R as directly
proportional to the intensity at the transducer acoustic window (Pr) by the equation
where D (cm) is the acoustic window diameter, W (cm) is the focal volume width at
depth R (cm), a is the absorption coefficient, and f is the applicator frequency. This
model approximates an ideal case for a spherically focused transducer in which the
acoustic intensity is uniform in the acoustic "cone" at any given plane from that
transducer aperture.
Since the model is based on power and not acoustic wave
interference, there are no grating lobes or undesirable pre-focal "hot spots" produced by
constructive interference.
Also unlike an acoustic model, the power is uniformly
distributed within the focal volume without distinct foci and the associated peaks and
troughs of heating.
Simulations using this model were performed on a PC (Micron,
Boise, ID) over the frequency bandwidth 1-2 MHz and focal depths of 6-20 cm into
tissue. The acoustic aperture was simulated for transducers with an f-number of 0.8-1.0.
49
D
WPf
Pt
- R
Fig. 3-1: Diagram of the area gain/axial attenuation model for determining the maximum effective
focal width for a single sonication using a phased array applicator.
The temperature elevations caused by the power deposition pattern were
calculated using a finite element analysis of the bioheat transfer equation (Pennes 1948)
and the necrosis threshold was calculated using the Sapareto-Dewey thermal dose model
(Sapareto and Dewey 1984) (see Appendix C for the details). This thermal dose model
has been experimentally verified in vivo with threshold levels for necrosis found to be in
the range of 30-240 equivalent minutes at 430 C (240 used as threshold in these
simulations) (Damianou et al. 1995). The simulated sonication time was 10 seconds and
the total power level was varied to adjust the peak simulated temperature.
As these simulations are trying to theoretically maximize coagulated tissue
volumes, a limitation must be set which distinguishes between a large coagulated volume
of deep seated tissue and a large coagulated tissue volume that is large only because of
the extensive near field heating. Clearly, the goal is to necrose deep seated tumor tissue
and not the pre-focal healthy tissue. Previous research has indicated that the threshold of
50
thermally induced pain occurs at about 450 C (Hynynen et al. 1990; Perez et al. 1993). To
thermally necrose tissue in a short time period, this threshold must be exceeded in the
target volume.
However, the volume of healthy tissue that exceeds 45 0C should be
minimized. Therefore, as an initial criterion for treatment, the simulations limited the
focal volume such that no tissue outside of 1 cm from the necrosis volume would surpass
the thermal pain threshold. This should result in a relatively pain-free treatment.
3.2.2 Array Element Design Given the Maximum Focal Volume
Given the maximum focal volume for a set tissue depth and available acoustic
aperture, the array parameters can be established. This is accomplished by using the well
known 6dB beam patterns of a focused piezoelement:
F
1 = KI (-)2
and
AF
2D
where w is the focal width, / is the focal length, X is the ultrasonic wavelength (about 1
mm at 1.5 MHz in water), F is the focal distance (radius of curvature) , D is the diameter
of the element, and K1 and K2 are constants which depend on the aperture angle of the
element (typically, K1 is between 8-14 and K2 is about 1) (Wells 1977). When the focal
width equation is applied using the dimensions of a single element of the spherical
sectioned array, the resulting width is approximately the limits of the array's ability to
electronically shift the focus away from the geometric focus of the array in the focal plane
51
(Wells 1977; Goss et al. 1996). By applying the width and length equations using the
array geometry as a whole, the resulting dimensions are approximately the dimensions of
individual foci formed close to the geometric focus.
Therefore, the dimensions of the
individual foci are set by the array configuration while the effective thermal focus of the
pattern of foci is set by the size and number of array elements. It is noteworthy that the
scanned or multiple focus patterns are always in the near field of the array but in the far
field of the array elements. Therefore, given the maximum theoretical focal width found
by using the area gain/axial attenuation model, one can determine the maximum size (and
minimum number) of elements in the array such that the array can scan foci throughout
the theoretical volume.
The array designed as an example for this paper was simulated using an acoustic
model based on the Rayleigh-Sommerfeld
superimposed point sources (Zemanek 1971).
integral over a set of geometrically
Temperature elevations were again
calculated using the Pennes bioheat transfer equation and the dose distributions were
calculated from a numerical integration of the Sapareto and Dewey model (Pennes 1948;
Sapareto and Dewey 1984). The spatial resolution was 0.25 mm in the transverse axis of
the array and 0.50 mm in the longitudinal axis. The temporal resolution was 0.02 s. The
phase distribution was calculated using the pseudoinverse or mode scanning approach
(Ebbini and Cain 1989; McGough et al. 1994).
3.2.3 Focal Spacing Simulations
The area gain/axial attenuation model is used to calculate the maximum focal volume
that an ideal array could coagulate in a single sonication. An actual array, however, can
52
not create a perfectly uniform power deposition pattern in the focal volume. Instead, the
uniform temperature and thermal dose elevation must be formed by the superposition of
individual foci or focal patterns. A study by Damianou (Damianou and Hynynen 1993)
investigated the distance between sequential sonications to ensure tissue coagulation
between foci using a fixed focus transducer.
Simulations for the current study
investigated the maximum distance possible to cause uniform thermal dose between
adjacent foci from a multiple focus pattern using a phased array.
This is done by
simulating a 3 x 3 cm 2 grid of foci in the focal plane of the array. The grid was formed by
superimposing the pressure magnitude squared fields of a single focus transducer of
varying frequency (1-2 MHz) and varying F-number (0.8-1.0). The power for the entire
grid was varied in a series of simulated 10 second sonications until the thermal dose at
the center of the grid was 5000 equivalent minutes at 430 C or until the peak temperature
of the sonication exceeded 1000 C. Fig. 3-2 contains a diagram of the simulated field.
3 cm (fixed)
1-5 mm (variable)
Fig. 3-2: Positioning of foci to determine the maximum spacing in a multiple focus pattern to
create a
uniform thermal dose.
53
3.3 Results
3.3.1 Area Gain/Axial Attenuation Model
A sample of the response using the area gain/axial attenuation model is found in
Fig. 3-3. In this case, the focal plane is located 7 cm into the tissue and the effective focal
width is set as 2 and 3 cm at a frequency of 1.5 MHz (tissue attenuation set as
5 Np/m/MHz). In each case, the lesion is localized well beneath the tissue surface as
shown in the thermal dose plots, but the temperature contours indicate that the wider
focal region causes extensive heating in the tissue between the transducer and the focus.
The predicted lesion also begins to extend more in the pre-focal region as compared to the
post-focal zone. This is not surprising since the larger focal width lowers the intensity
gain from the transducer beyond the focal plane. As the temperature rise in the near field
preludes the extension of the thermal dose necrosis, the pain threshold criteria is used to
limit the maximum focal width.
54
W= 2.0 cm
150
,_
,_
_
W= 3.0 cm
150
-650
100
100
50
50
-50
0
50
150
100
100
C/Q
50
-55C
.. -. 450 C
-50
150
0
50
240 Eq. Min
@ 430 C
50
-50
0
50
C
-50
0
50
Distance Across Axis (mm)
Fig. 3-3: Results of the area gain/axial attenuation model for a 1.5 MHz array with focal distance of
100 mm sonicating 70 mm into tissue. The top graphs correspond to the temperature contours at a
maximum temperature of 700 C in a 10 second sonication (contours at 450, 550, and 650 C). The lower
graphs correspond to the thermal dose contours of thermal coagulation (240 equivalent minutes at
430 C).
Fig. 3-4 and Fig. 3-5 are plots of the maximum focal width using the 450 C pain
threshold, criteria. Fig. 3-4 graphs the peak ratio of focal width to acoustic aperture
(W/D) against the tissue depth (R) with a variable peak temperature in the focal plane
(frequency set at 1.5 MHz). It is apparent from this graph that the lower peak temperature
can yield larger focal volumes. Fig. 3-5 plots the same type of curves but sets the peak
temperature to 700 C and varies the driving frequency. As expected, the lower frequency
can produce a larger lesion width than the higher frequencies, but the curve's response to
frequency is less sensitive than its response to peak temperature.
55
E 0.20
0
0.16
(0
-
600
-
800-
0.12
0
0.08
-
1000
0.04
.5
C.)
0 0.00
8
10
12
14
18
16
20
Depth of Focus (cm)
Fig. 3-4: Area gain/axial attenuation plots of maximum focal width to acoustic window width for a 10
second sonication at 1.5 MHz with varying peak sonication temperatures.
0
0.20
.........
1.0 M Hz
0.16
--
1.5 MHz
(0
0.12
.5
.9
0.08
.
MHz
N2.0
U
0
0.04
700 C peak
0
0.00
6
8
10
12
14
16
18
20
Depth of Focus (cm)
Fig. 3-5: Area gain/axial attenuation plots of maximum focal width to acoustic window width for a 10
second sonication with a peak temperature of 700 C and varying frequency.
56
3.3.2 Focal Spacing Analysis
The simulated response for a grid of 1.5 MHz foci with variable spacing is shown
in Fig. 3-6. At small interfocal spacing, the temperature and thermal dose are uniform
within the center of the volume. As that spacing becomes larger, the individual foci must
cause higher temperatures to ensure that the interfocal tissue is coagulated. The peak
focal temperatures needed for a given focal spacing to guarantee interfocal tissue
destruction are plotted in Fig. 3-7. Note that for very small focal spacing, the necessary
temperature elevation to coagulate tissue in a ten second sonication is approximately
600 C. As the spacing increases, the temperature also increases, although to a lesser
amount for lower frequencies (which have larger individual foci) and for smaller fnumbers (which also have larger individual foci). For all frequencies and for both fnumbers, the maximum focal spacing to create a relatively uniform dose is approximately
1.5 times the ultrasonic wavelength.
016000
;;
0
Focal Spacing
12000
2.0 mm
.
1.5 mm
c~
8000
-
1.0 mm
LU ~) 4000
W
00
0
0
5
10
15
20
Distance from Grid Center (mm)
Fig. 3-6: Example of a half cross-section, thermal dose levels as the distance between adjacent foci is
varied in the 3x3 cm 2 of foci. The individual foci are the superposition of the focus simulated of an fnumber 0.8, 1.5 MHz transducer. The center dose is 5000 equivalent minutes at 430 C.
57
Fixed f-number 1.0
a-
0
E
0)
7-
10C
90
Fixed frequency 1.5 MHz
2.0 MHz......
1.5 MHz
1.0MHz---- .
100 f-num 0.8
90 f-num 1.0 ---
80
.
60
60
-
1
-
80',
70
80
70
-,
2
1
4
3
Focal Spacing (mm)
2
3
4
Fig. 3-7: Maximum temperatures simulated for the grid of variable focal spacing. The center dose is
constant for all data points (5000 equivalent minutes at 430 C) while the f-number and frequency are
varied.
3.3.3 Design Example
The following is a sample design using the data from the area gain/axial
attenuation model, the array element beamwidths, and the focal spacing data. Suppose
that a liver treatment has the following constraints: tumor depth of 7 cm and maximum
acoustic aperture diameter of 12 cm. The frequency of the design should be selected as
low as possible since it can yield larger focal widths with wider interfocal spacing.
However, the lower frequencies are also more likely to cause thermally significant
cavitation (approximately a linear relationship of threshold P = 0.6+5.3*f where P is the
pressure in MPa and f is the frequency in MHz) (Hynynen 1991). For this design the
frequency was chosen as 1.1 MHz. This yields a maximum focal width of 1.9 cm (0.16
W/D from Fig. 3-5 with D = 12 cm to match the acoustic window and a peak temperature
58
of 70 C). The array could be built out of a 0.83 f-number transducer. Using the 6 dB
beamwidth equations, the array element size must be less than 7.2 mm in diameter (the
diameter of the focal region should be within the 6 dB beamwidth of the individual array
elements) leading to an array of around 220 elements. To treat a 1.9 cm wide focal region
with a peak temperature of 700 C, a focal spacing of 3 mm is necessary. Thus the focal
grid would consist of 49 foci. This pattern can be produced using a phase generation
technique such as the pseudoinverse or phase rotation in conjunction with temporal field
multiplexing.
An acoustic simulation shows the difference between the simplified theoretical
model used above and a practical array design. Fig. 3-8 demonstrates the simulated
heating and necrosis volume from a 256 element, 1.1 MHz array used to treat a 1.0 x 1.0
2
cm 2 focal area. This array has element sizes of 6.5x6.5 mm2. A grid of 25 foci are
formed by temporally multiplexing between multiple focus patterns. While an array this
size can theoretically necrose a 1 cm diameter volume without causing any "hot spots" or
excessive near field heating, the simulated array does not uniformly distribute power in
the near field and "hot spots" are formed. There are extensions of temperature elevations
extending from the lesion towards the array.
Therefore, an improvement in phase
distribution driving signals and/or an increase in the number of elements in the array are
still necessary to make this treatment effective.
59
130
130,
120
120
110
110
100
100
90
90
80
80
-20
70
70
60
60
-40
-40
50
-40
50
40
20
0
0
-20
x
(mm)
20
40
0
-20
x (mm)
20
40
_
0
-50
50
diag (mm)
Fig. 3-8: Temperature and thermal dose contours simulated for a 256 element, 1.1 MHz phased array
with 12 cm diameter and 10 cm radius of curvature. The contours are the predicted lesions (black
lines) and the 450 C temperature contour (gray lines) of the cross axis of transducer in the focal plane
(left), along the axis of the transducer perpendicular to "square" in focal plane (middle), and along
the axis of the transducer diagonal to the "square" in focal plane (right). The peak temperature was
70* C in a 10 second sonication.
3.4 Discussion
Previous research has noted that using a phased array for thermal ablation can
decrease the treatment time of large tumors by 60-75% (Wan et al. 1996; Fan and
Hynynen 1996b). Phased arrays are capable of this since they can scan a single focus
very rapidly and can create multiple focus patterns. They offer the control to shape the
acoustic deposition pattern to optimize the temperature and thermal dose response of the
tissue thereby increasing the effective focal volume of tissue that can be treated in a
single sonication. The area gain/axial attenuation model can be used as a first order tool
to design spherical sectioned phased arrays for treating deep seated tissue. The model can
indicate the maximum treatment volumes for any acoustic aperture and tissue depth.
Using the maximum values, the designer can determine the minimum number (and
maximum size) of the array elements necessary for the treatment application.
It is noteworthy that the maximum necrosis volume is also a function of its
limiting criteria (in this case the limitation was set as limiting the pain threshold
60
temperature to within 1 cm of the lesion). A different criterion which either changes the
threshold temperature in non-diseased tissue or alters the size of the thermal boundary
around the lesion could yield larger or smaller volumes. In the case of liver treatment,
where the hope is to perform the treatment in a single breath-hold period, the pain
criterion was deemed appropriate.
Other treatments could use a different requirement
depending on the use of anesthesia and sonication duration.
Using the criterion of this
study, the maximum tissue necrosis volume for depths of 6-20 cm is less than 2 cm wide
and 3 cm long.
Even with a different limiting criterion, the area gain/axial attenuation technique
indicates some important parameters to improve the treatment of large tissue volumes. It
indicates that the f-number for the array should be less than 1.0 and that the frequency
should be reduced as long as the cavitation threshold is avoided. However, the change of
maximum focal width when frequency is varied (1-2 MHz) is only about 20% for depths
of 6-20 cm. Most importantly, the field should be optimized to decrease the peak and
average temperatures in the focal region to avoid excess heating in the near field.
The acoustic simulations indicate that the current technology for generating phase
and power distributions on the array surface needs more work to approach the idealized
situation. Non-uniform near field heating can create potential hot spots which at worst
could generate non-repairable tissue damage in the pre-focal tissue. This illustrates the
need for temperature sensitive feedback methods obtained from ultrasound, CT, or MRI.
Despite their non-ideal nature as compared to the area gain/axial attenuation
model, phased arrays offer much potential towards the fast treatment of large tissue
volumes. The spatial dimensions between adjacent foci in a multiple focus pattern can be
61
50% larger than those of sequential sonications of a single focus transducer (Damianou et
al. 1995), thus helping to create a more optimized power deposition with less wasted
energy in tissue that has already been coagulated.
Lastly, several researchers have indicated that a major advantage of the spherical
sectioned phased array is its ability to focus energy such that acoustic barriers such as
bones can be avoided (McGough et al. 1996; Botros et al. 1997). By applying the area
gain/axial attenuation technique to only the acoustic window area (as opposed to the array
aperture) the model can still be used to estimate maximum necrosis volumes.
62
4. ARRAY CONSTRUCTION AND ARRAY MATERIALS
4.1 Introduction
4.1.1 Array Requirements
This study investigated the materials and construction techniques for creating a
large scale, 2-D, therapeutic ultrasound phased array. The array had to be compatible for
use in an MR scanner, able to withstand extended periods submersed in water, generate
acoustic power levels sufficient for clinical tumor treatment, and minimize inter-element
coupling. This study will investigate the maximum power, efficiency, and inter-element
coupling for cut PZT-4 piezoelements and for 1-3 composite piezoelements manufactured
by two companies. It will then present the results of a constructed 256 element array.
4.1.2 Current State of Array Construction
Two dimensional diagnostic ultrasound arrays are usually constructed from a
single PZT piezoelement.
The technique, known as "dice-and-fill," cuts through an
existing crystal and then fills the spaces (called kerfs) with an ideally non-mechanically
coupling material (typically silicone, epoxy, or polymer). The cutting decreases the interelement coupling and the kerf material helps ensure that the geometry between elements
is kept intact. The geometry is kept rigid in most diagnostic arrays because they are
backed by a solid acoustic damping material.
On the transmitting side of the
piezoelement, the arrays may contain either a solid or diced quarter wave acoustic
matching layer(s) to improve ultrasound transmission, reception, and bandwidth.
63
Electrical connections are often made with conductive epoxy or paint (typically silver). A
conductive coating at the transmission side of the array acts as a ground plane for the
electronic signals.
Therapeutic transducers have a basic requirement that makes their construction
different from diagnostic devices: they must generate high levels of power for extended
periods of time. This includes continuous wave signals that last for durations up to hours
in the case of hyperthermia treatments. The implications of this requirement change the
design of the arrays significantly. First, the backing material of an ultrasound transducer
must be chosen to minimize transmission losses. This is most easily done by backing the
transducer with a very low impedance medium (air) (Wells 1977; Hynynen 1990).
Second, since a piezoelement is usually driven at its resonant monofrequency to
maximize its electroacoustic efficiency, the importance of a broadband matching layer is
decreased. It has been shown for several diced PZT elements that sufficient ultrasound
transmission can be obtained without a matching layer. Third, extreme care must be
taken to only use piezoelements whose dimensions tend to create single vibrational
modes (De Silets 1978; Challande 1990). While this requirement is also important for
diagnostic elements, it is critical for therapeutic elements where the non-transmitted
vibrational modes can lead directly to excess heating in the piezoelectric material and
failure of the transducer. Lastly, the current and voltage requirements are higher for the
electrodes of therapeutic devices. This makes the use of silver epoxy or conductive paint
less appealing in creating a conductive ground plane on the transmission side of the array.
As a part of this study, four prototype arrays were constructed using PZT and the
traditional "dice-and-fill" techniques.
64
4.1.3 Composite Materials
In the 1980's a series of papers from the Materials Research Laboratory of Penn
State were published describing a new type of piezoelectric material called 1-3 composite
(Shrout et al. 1980; Gururaja et al. 1980; Gururaja et al. 1984; Gururaja et al. 1985a;
Gururaja et al. 1985b; Gururaja et al. 1985c). This material consisted of small, embedded
piezoelement rods (such as PZT) in a polymer or epoxy base (see Fig. 4-1).
The
placement of these piezoelement pieces was two dimensionally periodic with varying fill
ratios of polymer to piezoelement. The use of this new type of material was auspicious
for several reasons. First, after placing the rods in the polymer the transducer can be
molded to any shape and size. This is an advantage compared to the mechanical limits of
grinding a PZT pseudocrystal. Second, the element division of a composite material does
not require the material to be diced, but only that the electrode be etched in a particular
pattern. Difficult patterns can therefore be created without costly machining. Third, the
polymer/piezoelement combination had a lower acoustic impedance to more closely
match a liquid coupling medium.
Lastly, the small piezoelements embedded in the
polymer contain appropriate dimensions to promote a single vibrational mode. They can
then be combined to produce element sizes that have similar width and thickness
dimensions that previously could not be used due to their low efficiency.
Recently, two companies have begun to advertise 1-3 composite piezoelectric
materials for high power applications (Materials Systems, Littleton, MA and Imasonic
SA, Besancon, France). This chapter will present quantitative measurements of these
new materials to investigate their use in therapeutic ultrasound arrays. The results will be
compared to diced PZT test arrays. In addition, the benefits of an acoustic matching layer
65
for therapeutic arrays will be quantitatively measured for two sample arrays. Lastly, this
chapter will present the results of a 256 element array designed for therapy (see Chapter 6
for therapeutic results).
Fig. 4-1: Diagram of a 1-3 composite piezoelectric. The rods represent the PZT and the fill between
the rods (not shown) is typically a polymer.
4.2 Methods and Materials
4.2.1 Acoustic Efficiency Measurements
All electroacoustic efficiency measurements were made using a radiation force
technique (Stewart 1982). Electrical power measurements were made for individual array
elements using either an HP 438A Power Meter (Hewlett Packard, Englewood, CO) or
the custom built power meters on the ultrasound driving system (see Chapter 2 and
Appendix A).
Efficiency measurements for powering the entire array simultaneously
used the custom built power meters for each individual element attached to the driving
system.
66
4.2.2 Maximum Power Measurements
The maximum acoustic power was measured by increasing the electrical power to
an array element and measuring the acoustic output power until it peaked and began to
drop. Efficiency measurements were then recorded at lower powers to measure the effect
on the transducer elements. Peak power measurements were performed on all of the
composite test arrays except for the therapeutic 256 element array.
4.2.3 Inter-element Coupling Measurements
Inter-element coupling was tested by measuring the acoustic output power when
adjacent elements were electrically driven in-phase and out-of-phase. Two elements were
electrically driven using separate but synchronized amplifiers with power levels such that
no irreversible damage was caused to either element. The electrical phase of one of the
amplifiers was shifted until the voltage on its transducer element was either completely
coherent (in-phase) or incoherent (out-of-phase with a 180 degree phase shift) with the
neighboring element. For each material, individual electrical power measurements were
made using two Hewlett Packard 438A Power Meters. The radiation force technique was
used to measure the acoustic power. Multiple pairs of elements were tested for each
array.
In addition, measurements of six adjacent elements of the Imasonic prototype
array and all 256 elements of the therapy array were taken while being driven
simultaneously. In these cases the power measurement and driving signals were provided
by the in-house built ultrasound driving system.
67
4.2.4 Acoustic Field Simulation and Measurement
The ultrasound fields of the 256 element array were simulated using the RayleighSommerfeld integral over a set of geometrically superimposed point sources as described
by Zemanek (Zemanek 1971). The acoustic vibration on the surface was modeled as
uniform. All calculations were performed on a dual Pentium II processor 300 MHz PC
(Micron, Boise, ID).
Automated stepper motors (Velmex, Bloomfield, NY) and a 0.075 mm
hydrophone (Precision Acoustic, Dorset, England) were used to scan the 256 element
array in a degassed water bath. Rubber matting was placed around the sides of the bath
and hydrophone to reduce reflections. The spatial sampling was equal to or less than
0.2 mm.
4.2.5 "Dice-and-Fill" Arrays
(*IMPORTANT NOTE: Some of this work was completed by the author while
other parts of the work were completed by Patrick Lopath prior to the beginning of this
thesis. All of the results were placed in this thesis since they were non-published but help
complete the series of tests performed as a part of this thesis.
The section subtitles
marked with a star (*) indicate that the work was mainly performed by Lopath.)
The dice and fill technique was studied by constructing small test arrays (not used
for therapy) and prototype arrays (such as the arrays described in other chapters of this
thesis used in vivo or ex vivo experiments). The variable acoustic measurements were
predominantly made on the test arrays while the prototype arrays were used to indicate
the most robust techniques for large array construction.
68
4.2.5.1
"Dice and Fill" Test Arrays
4.2.5.1.1 Piezoelectric Material for "Dice and Fill" Arrays
Two types of non-composite piezoelectric materials were tested for an array of
small square elements: lead zirconate titanite (PZT-4) and lithium niobate (LiNO). The
materials were flat and the dimensions were varied from a width-to-thickness ratio of
greater than 5.0 in both dimensions to a width-to-thickness ratio of 2.6 (thickness is set by
the resonant frequency wavelength). It was the intention that these materials could be
used to form large non-planar arrays by either dicing a large curved PZT-4 pseudocrystal
into smaller elements or positioning small flat pieces of LiNO on a spherical shell.
4.2.5.1.2 Kerf Adhesive for "Dice and Fill" Arrays
To join adjacent diced elements, five different adhesives from four companies
were studied. These are listed in Table 4-1. Manual shear tests of structural integrity
were used to rule out some blends of the adhesives.
Test arrays (typically just two
elements) were created for those materials which passed the manual shear test. Water
resistance was tested by submersing the test arrays overnight. Acoustic coupling tests
were used to find the best adhesives and to optimize mixing ratios for the multiple
component adhesives.
69
Table 4-1: Kerf adhesives tested for diced transducer arrays.
Adhesive
1
2
3
4
5
Company
Loctite
Dow
MERECO
Loctite
Dow
Dow
Material Names
V-1022
3140
RTV-1
Permatex 65A
D.E.R. 732
D.E.R. 331
Description
two part rubber potting compound
Low viscosity silicone
High viscosity silicone
glass sealant
epoxy resin softener
epoxy resin
Dow
DEH 24
hardener
4.2.5.2 "Dice and Fill" Prototype Arrays
Using the dice and fill techniques found in the test arrays, four prototype arrays
were constructed.
A description of the arrays is found in Table 4-2. They were all
Three types of cable have been tested
designed in the spherical sectioned geometry.
using the prototype therapy arrays. Some of the cable properties are listed in Table 4-3.
Connections between the piezoelectric material and the coaxial cables using conductive
epoxy, solder, paints, and pogo pins* were investigated.
Table 4-2: Prototype therapy arrays.
Elements
Electrode
Kerf Adhesive
Number of Element
2
Connection
Size (cm )
16
4
16
4
64
1
and Cable
Dow D.E.R. 331
(see material 6 in
table)
Dow D.E.R. 331
(see material 6 in
table)
Loctite V-1022
Silver
epoxy
paint/silver
Cu/Au
electrode/solder
5 m,
Belden
28 AWG
5 m,
Belden
28 AWG
28 AWG
1)
Pogo pins
5 m,
2)
Cu/Au electrode
Belden
and solder
76
1
Cu/Au
Loctite V-1022
I electrode/solder
70
5 m,
Tensolite
30 AWG
Table 4-3: Types of coaxial cable used tested in therapeutic arrays.
Manufacturer
Belden
Tensolite
Precision Interconnect
Part Number
8700010
30830/41119X-1(LD)
155-0205-7NN
AWG
28
32
34
Capacitance (pF/m)
181
78
115
4.2.6 1-3 Composite Materials
4.2.6.1 1-3 Composite from Material Systems
Four types of composite piezoelectric materials were obtained from Material Systems,
Inc. (Littleton, MA, USA).
These materials were all 1-3 composites using PZT-4 at a
resonant frequency of 500kHz. The electrode consisted of silver epoxy (the company did
not manufacture solid electrodes). The material data are found in Table 4-4.
Table 4-4: 1-3 Piezocomposite obtained from Material Systems, Inc.
Material ID
MS1
MS2
MS3
MS4
Fill Ratio %
30
30
40
50
Polymer Fill Matrix
Shore D80 PU
Voided shore D80 PU
Shore D80 PU
Voided shore D80 PU
W/H
Freq.
Ratio
(MHz)
0.41
0.41
0.26
0.46
Two test arrays were constructed from each composite material.
0.5
0.5
0.5
0.6
Each array
consisted of a 1 cm 2 square transducer which was electrically subdivided into four
0.5 x 0.5 cm 2 elements by etching one side of the epoxy into quadrants. The elements
were joined electrically to 28 AWG, 1 m Belden coaxial cable using silver epoxy. The
transducer was then positioned in an air backed transducer case using a silicone sealant.
The resonant frequency for each element was determined by choosing its maximum real
impedance as measured on a Hewlett Packard 4193A impedance meter. The elements
71
were electrically matched using simple LC circuitry such that their impedance was 50 Q
at the frequency of interest.
4.2.6.2 1-3 Composite from Imasonic
4.2.6.2.1 9 Element Test Array
A nine element test array was obtained from Imasonic (Besancon, France). The
array consisted of a 3-by-3 grid of square elements with 0.5 cm sides yielding an aperture
of 1.5 x 1.5 cm2 . The array had been placed in an air backed casing by the manufacturer
with metal electrodes attached to the holder for electrical connection.
The type of
connection to the composite material (solder, epoxy, etc.) was not disclosed by the
company. The front of the transducer had an undisclosed, proprietary acoustic matching
layer.
The test array was connected to either 1 m Belden or 5 m Tensolite cable using
solder to the casing electrodes. The resonant frequency for the array was determined by
finding the maximum real impedance for each of its elements as measured on a Hewlett
Packard 4193A impedance meter. The elements were electrically matched using simple
LC circuitry such that their impedance was 50 Q at the frequency of interest (0.88 MHz).
The elements were later matched at 1.0 MHz which was the specified frequency
requested to the manufacturer.
4.2.6.2.2 256 Therapy Array
A 256 element spherical sectioned array was designed to treat a 3 cm 3 volume in a
single sonication. The frequency was chosen to be 1.1 MHz to avoid the cavitation
72
threshold of low frequency ultrasound and the increased attenuation of high frequency
ultrasound.
The element size was determined by maximizing the surface area of the
transducer while keeping the element projections of equal area.
The array geometry
(element size and frequency) were simulated to ensure that the array could adequately
shift a single focus within the target volume without grating lobes greater than 10% of the
main lobe.
The 256 element array was then constructed from a spherically curved 1-3
piezocomposite shell (Imasonic, Besancon, France) with a 10 cm radius of curvature and
Although the array geometry has been titled "spherical sectioned"
12 cm diameter.
(Ebbini and Cain 1991a), the 1-3 composite material was not "sectioned" by cutting
through the material.
The array elements were etched on the convex electrode of the
array using a diamond wire saw in a pattern whose planar projection is a grid of
s (see Fig. 4-2). Unlike the prototype array, the front of the
0.65x0.65 CM 2 squares(eFi.42.Ulkthprttparahefotfte
transducer did not include a quarter wave acoustic matching layer. On the back of the
array, one end of a small strip of silver foil was directly soldered to each element
electrode. Each foil was connected to a 7 m, 34 AWG, magnet compatible, microcoaxial
cable (Precision Interconnect, Portland, OR).
including
the
DC capacitance
piezoelement/cable
and
Electrical measurements of the cable
conductivity
were
used
to
estimate
the
impedance using a single stage "L" transmission line model
(Anderson 1995). Comparisons to the measured impedance (using an HP 4193A Vector
Impedance Meter, Hewlett Packard, Englewood, CO) of the cabled transducer were made
to verify the model and estimate efficiency losses in the cable.
The electroacoustic
efficiency of the array's piezocomposite was measured as connected with the 7 m, 34
73
AWG coaxial cable for the entire array and with 1 m, 28 AWG cable for two sample
elements. A photograph of the completed array is found in Fig. 4-3.
Fig. 4-2: Planar projection of array elements. The elements are shaded and the small triangular
spaces on the array edges were used for ground connections and were not powered.
74
Fig. 4-3: Photographs of the 256 element array.
75
4.3 Results
4.3.1 "Dice-and-Fill" Tests
4.3.1.1 Quantitative Results from the "Dice and Fill" Test Arrays
4.3.1.1.1 Non-Composite Materials Tested for "Dice and Fill" Arrays
Table 4-5 and Table 4-6 contain the electroacoustic efficiency data for the PZT-4
and LiNO piezoelectric materials. It was found that for a width-to-thickness ratio greater
than 5 that the both materials had acceptable efficiencies (66% for PZT-4 and 49% for
LiNO).
As the ratio dropped to 2.6 for both dimensions, the efficiencies dropped
significantly. The materials offered similar efficiencies although LiNO was somewhat
higher.
Table 4-5: Square element test array results for PZT-4.
X dimension (W/T)
>5
>5
>2.6
Y dimension (W/T)
>5
2.6
2.6
Efficiency (%)
66
40
24
Table 4-6: Square element test array results for lithium niobate.
X dimension (W/T)
>5
>5
>2.6
Y dimension (WIT)
>5
2.6
2.6
Efficiency (%)
76
58
49
4.3.1.1.2 Kerf Adhesives for "Dice and Fill" Arrays
Only two of the five different types of adhesives passed a manual manipulation
test.
These included the Loctite silicone (V-1022) and Dow epoxy combination.
After
being submersed, both materials experienced a weight gain of less than 1%. The initial
76
efficiency of the test arrays was 78% (silicone) and 74% (epoxy). The efficiency of the
array with epoxy kerf dropped to 62% after 10 minutes of 20 W power while the
efficiency of the array with a silicone kerf did not drop under the same conditions.
Three test arrays, each with two elements, were prepared. The first array did not
have a diced kerf but only an etched electrode to divide the elements. The second and
third arrays used epoxy and silicone to fill a diamond wire cut kerf (0.3 mm width). The
results of interelement coupling tests with in phase and out of phase powering are found
in Table 4-7.
Table 4-7: Interelement coupling dependence on kerf fill at an average acoustic power of 8 W.
Kerf Type
Scoured electrode
Silicone
Epoxy
In Phase Efficiency (%)
64.3
67.3
62.9
Out of Phase Efficiency (%)
57.9
66.4
63.2
4.3.1.2 Qualitative Results from the "Dice and Fill" Prototype Arrays
4.3.1.2.1 "Dice and Fill" Array Cable Type
The three types of coaxial cable were found appropriate for different types of
arrays. First, a 28 AWG cable (Belden) was found appropriate for arrays with a small
number of large elements since it introduced a small insertion loss and its outer insulation
is easy to encapsulate at the array holder. It can carry over 50 W of RF energy into a large
element. Its high capacitance and large diameter, however, make it inappropriate for
arrays with many elements.
The 30 AWG cable (Tensolite) was tested for small
elements. Its low capacitance (79 pF/m) and low loss (<0.5
/m) allows the elements to
be easily matched, but its stiffness and size contraindicate its use for an array with a large
77
number of elements. In addition, it is very difficult to water-proof (pot) the cables into a
holder due to the PVC coating. Lastly, a 34 AWG cable (Precision Interconnect) was
tested for an array with a large number of elements. The cable has low capacitance (98
pF/m) and relatively small resistance (1 fl/m for 34 AWG). It can withstand continuous
RF power of over 10 W as tested for several piezoelements from 1-2 MHz. It is flexible
and small for dense arrays.
The high resistance of the cable, however, can cause
excessive insertion loss (see composite materials section).
4.3.1.2.2 Array Grounding for "Dice and Fill" Arrays
All of the diced arrays experienced problems with their water facing ground
planes leading to elements with poor ground connections. First, tests of silver paint and
silver epoxy in the first 16 element array failed in the long term due to exposure to water.
The second 16 element array contained a copper/gold electrode and was professionally
diced such that small connecting tabs of PZT connected the elements at the corners.
Unfortunately, under high power, these tabs broke and cracked the electrode.
Solder
bridges were used to repair the array. The 64 and 76 element arrays used solder joints at
the corners of the elements to connect the ground plane.
This works well for large
elements but it would degrade array performance as the array elements become smaller
and more of their surface area is covered by solder. A technique to deposit a robust
metallic ground plane over the kerf's silicone or epoxy was not found although it was
attempted.
4.3.1.2.3 Cable Connections to "Dice and Fill" Arrays
78
The signal cables to the first 16 element array were connected with silver epoxy
(Bipax, Tra-con, Medford, MA).
The connections worked well for a number of
experiments but eventually disjoined from the elements over several sonications. Spring
loaded pogo pins were tested* in the 64 element array. There was no difference in
transducer efficiency measured between using pogo pins or a solder joint. Unfortunately,
the spring loading of the pogo pins tended to push the array elements out of alignment
(this array used silicone kerfs). Low temperature solder was found to be the most reliable
connector (as seen in the second 16 element array and the 76 element array).
4.3.1.2.4 Kerf Adhesives for the "Dice and Fill" Arrays
Neither the epoxy nor silicone kerf materials was found to offer a long term,
robust solution. First, although the silicone was able to water proof the prototype arrays,
it did not give adequate structural support for the larger arrays (64 and 76 elements).
Second, the epoxy filled kerf gave good temporary support to the 16 element arrays, but
over time the epoxy became brittle and broke away from the PZT elements causing the
array to leak.
4.3.2 1-3 Composite Materials Tested for Array Construction
4.3.2.1 1-3 Composite from Material Systems
The electroacoustic efficiencies of the materials supplied by Material Systems are
tabulated in Table 4-8. These efficiencies were measured using approximately a 2 W
input power. The 40% fill material (MS3) had by far the best efficiency. This is most
likely explained by the smaller width-to-thickness ratio of the pillars. The efficiency
79
dropped, however, as power was increased to the elements.
Fig. 4-4 shows the
relationship between electrical input power levels and the output efficiency for material
MS3.
Table 4-8: Efficiency data for composites from Material Systems, Inc.
Material
MS1
MS2
MS3
MS4
Average Electroacoustical Efficiency (%)
8.6
13.7
50.9
17.0
Range of Efficiencies (%)
3.9-11.2
11.7-16.3
43.6-55.6
12.8-19.1
60
50 -
40
30
0
30
02 20
0
10
0.OOE+00
2.00E+00
4.OOE+00
8.OOE+00
6.OOE+00
1.OOE+01
1.20E+01
Electrical Input Power (W)
Fig. 4-4: Electroacoustic efficiency as a function of input electrical power for MS3 (40% fill
material). The points correspond to measurements taken from the eight elements on the two arrays.
Two
factors
limited
the
maximum
power
from
Material
System's
piezocomposites: electrode failure and decreased electroacoustic efficiency. In the case
of decreased efficiency, the elements could recover.
The electrode failure caused
complete element destruction. Table 4-9 lists the maximum powers achieved from the
four materials before the electrode failed.
80
Table 4-9: Maximum acoustic power from composites manufactured by Material Systems, Inc.
Material
Average
Maximum Power
Range of Maximum
Power (W/cm 2)
Efficiency at
Maximum Power
(W/cm 2)
(%)
5.4
10.0
38.8
17.8
0.88-1.32
2.16-3.04
8.0-12.3
5.6-7.1
0.93
2.76
10.4
6.2
MS1
MS2
MS3
MS4
Since the maximum output power levels were so low for materials MS 1 and MS2,
reliable acoustic coupling measurements could not be made (low power coupling
appeared to be negligible). Measurements for materials MS3 ad MS4 are found in Table
4-10.
Table 4-10: Inter-element coupling for Material Systems, Inc.
Material
MS3
MS4
In Phase
Out of Phase
% Drop in Output
Efficiency (%)
Efficiency (%)
Power
53.0± 3.3
17.5±2.4
50.4±3.4
15.0±0.8
4.5
14.2
4.3.2.2 1-3 Composite from Imasonic
4.3.2.2.1 Nine Element Test Array
Table 4-11 contains the efficiency measurements at 0.88 MHz (frequency of
maximum real impedance) and at 1.00 MHz (specified frequency) for approximately 8
W/cm2 acoustic output power applied to the Imasonic nine element test array using 1 m
Belden coaxial cable.
Table 4-12 contains the efficiency measurements when the
elements were driven through 5 m Tensolite coaxial cable.
After driving elements
numbered 2 to 4 continuously at 8-12 W/cm 2 for more than an hour, there was no
noticeable change in electroacoustic efficiency.
81
Table 4-11: Electroacoustic efficiencies at approximately 8 W/cm2 acoustic power output through 1 m
long Belden coaxial cable.
Efficiency 0.88 MHz
Element #
1
2
3
4
5
6
7
8
9
Average
Std Dev
Efficiency 1.00 MHz
61.2
57.6
57.3
61.4
56.5
54.9
53.7
54.0
53.0
56.6
2.92
66.3
64.5
66.4
66.4
68.2
68.6
67.6
65.4
67.7
66.8
1.27
2
Table 4-12: Electroacoustic efficiencies at approximately 8 W/cm acoustic power output through 5 m
Tensolite coaxial cable.
Element #
Efficiency 1.00 MHz
7
8
9
49.8
41.5
47.2
Average
46.1
Std Dev
3.5
Maximum power of the Imasonic prototype array was measured for three elements
using the Belden cable and for 5 elements using the Tensolite cable. These data are found
in Table 4-13 and Table 4-14. The maximum power, however, did not cause complete
2
element failure. For example, after exceeding its maximum power level of 16.5 W/cm
element 2 could still repeatedly reach an output power level of 14 W/cm 2 . For all of the
elements, the efficiencies at lower power levels dropped less than 3% after slightly
exceeding their maximum power. If the maximum power was substantially exceeded
(greater than 10% over its maximum) then element failure occurred and efficiencies
dropped up to half of their original values.
82
Table 4-13: Maximum acoustic power levels for three transducer elements using 1 m Belden cable.
Element
Maximum Acoustic Power (W/cm2)
2
3
4
Average
Std Dev
16.56
19.68
18.04
18.1
1.56
Table 4-14: Maximum acoustic power levels for three transducer elements using 5 m Tensolite cable.
Element
5
6
7
8
9
Average
Std Dev
Maximum Acoustic Power (W/cm 2)
17.72
18.60
17.64
17.24
15.64
17.4
0.97
The inter-element coupling results for the Imasonic prototype array are found in
Table 4-15 and Table 4-16. The average decrease in output power due to two elements
driven with opposite phases was 13% but ranged up to 18%. When six elements are
driven such that their phases alternate between 0' and 180' then the output power drops
between 18-32% depending on the input power level. The high coupling most likely is
attributable to the solid matching layer on the front of the transducer.
Table 4-15: Coupling measurements of adjacent elements at 1 MHz using 1 m Belden cable.
Elements
1&2
3&6
4&5
5&8
7&8
8&9
6&9
5&8
Average
Std Dev
In Phase Eff. (%)
59
60
42
42
47
48
49
49
49.5
6.3
Out of Phase Eff. (%)
53
56
38
39
39
39
40
40
43
6.7
83
% Decrease of Power
10.2
6.7
9.5
7.1
17.0
18.8
18.4
18.4
13.3
5.0
Table 4-16: Coupling measurements of 6 adjacent elements (elements 4-9) at 1 MHz using 5 m
Tensolite cable.
Input Power Level
(W)
6
12
24
36
In Phase Efficiency
(%)
56
53
50
45
Out of Phase Efficiency
(%)
46
41
34
32
% Drop in Output
Power
18
23
32
29
4.3.2.2.2 256 Element Array
The electroacoustic efficiency of twelve elements of the 256 element array
connected to the 7 m Precision Interconnect cable were tested individually using the ENI
amplifier and HP power meter. The results are found in Table 4-17. Two of these
elements were also tested using the 1 m Belden cable (see Table 4-18).
Table 4-17: Electroacoustic efficiency for elements of the 256 element array connected with 7 m
Precision Interconnect cable.
Acoustic Power (W)
3.07
2.99
2.84
2.82
3.04
3.33
3.26
3.45
3.37
3.68
3.33
4.35
3.29
0.41
Element Number Electrical Power (W)
8.90
117
8.20
118
7.12
119
7.90
120
7.50
121
7.80
122
7.70
123
8.36
124
8.40
125
8.80
126
8.50
127
10.3
128
8.29
Averages
0.82
Std. Dev.
84
Efficiency
34.52
36.52
39.94
35.75
40.54
42.7
42.35
41.23
40.11
41.78
39.21
42.18
39.7
2.73
Table 4-18: Electroacoustic efficiency measurements from two elements of the 256 element array
through 1 m Belden cable.
Acoustic Power (W)
1.93
1.80
1.87
Element Number Electrical Power (W)
2.97
117
2.88
118
2.93
Average
Efficiency
64.97
62.55
63.8
The "L" electrical model estimated that 55% of the electrical power delivered to
the therapeutic transducer would be lost in the 34 AWG microcoaxial cable (see Fig. 4-5
for a comparison of the element impedance and Fig. 4-6 for the estimated loss in the
cable).
20
5000
Element Impedance
without Cable
4000.
0
3000.
-20.
W
2000
-
0
0.8
0.9
1
-80
0.8
1.2
1.1
Frequency
x
1.1
1.2
1
X 106
Frequency
0.9
1
1.1
1.2
x 106
Frequency
0.9
1.1
1.2
1
X 106
Frequency
-75
200
-80
180.
a-
-85
160
1401
0.8
0.9
1.1
1
Frequency
Measured Element
Impedance with
Cable
0.9
106
220
Predicted Element
Impedance with
Cable
-40
-60
1000
0)
1.2
-90
0.8
x 106
220
-70
200
-75
180.
-80
160.
-85
.a
1401
0.8
0.9
1
1.2
1.1
Frequency
x
16
-90
0.8
Fig. 4-5: Measured and estimate electrical impedances using the "L" transmission line model for the
1-3 composite piezoelements (magnitude on left and phase on right).
85
1.0
-
-
0.8
o
0 0
.2.c0.6.
-. J
0.4.
-5
0.2.
o
0.
0.8
1.1
1.0
0.9
1.2
Frequency (MHz)
Fig. 4-6: Estimated power loss in the 8 m, 34 AWG cable as calculated using a single stage "L"
coaxial model.
The maximum acoustic power measured for the entire therapy array was over 500
W although the elements of this array were not driven until they failed. Individually,
powers up to 4 W per element were measured.
The measured interelement coupling for the 256 element array is found in Table
4-19.
During the measurements, the individual pairs were powered at about 1.2 W
acoustic power per element and the entire array was driven at 250 W acoustic power (1 W
per element).
Table 4-19: Interelement coupling for the 256 element array.
Elements
119&120
119&137
121&136 (non adjacent)
All 256 elements
In Phase Efficiency (%)
36.5
35.8
33.5
32.4
86
Out of Phase Efficiency (%)
34.8
34.1
33.7
30.0
4.3.3 Hydrophone Scans of the 256 Element Array
4.3.3.1 Single Focus Shifting
Fig. 4-7 shows the simulated and hydrophone measured intensity pattern for a
single focus that is shifted off the array axis in the focal plane.
Fig. 4-8 indicates the
simulated and experimentally scanned off axis focal shifting limits of the array. Fig. 4-9
plots the normalized hydrophone measured focal pressure magnitude squared as the focus
is electronically shifted along the array axis between 8 and 12 cm from the array.
E 10
E
0
-10
E
E 10
g4
0
0
*
-10
-10
.
10
X-axis (mm)
-10
10
X-axis (mm)
Fig. 4-7: Simulated (left) and hydrophone measured (right) intensity scans for a single focus at the
geometrical focus of the array (top) and shifted off axis 7 mm in the focal plane (bottom)
87
Experimental
Simulated
10
-- 10,
E
*
9 0
e
a
0
0
-1
-10
0
-10
-10
10
10'
-10
0"
0
-10
.
.
10
0
10
0
10
.
-10
10
X-axis (mm)
-10
0
0
X-axis (mm)
Fig. 4-8: Limits of off axis focal shifting in the focal plane (simulated on left and experimentally
measured on right).
1.0
3.8
-9
-4
3.6
3.4
E
0
z
3.2
3.0
6( )
100
80
Array Axis (mm)
120
Fig. 4-9: Normalized intensity hydrophone measurement of a single focus scanned along the axis of
the array.
88
4.3.3.2 Multiple Focus Fields
To demonstrate the ability to create more complex focal patterns 16 and 25 focus
fields were generated using the pseudoinverse technique (Fig. 4-10) (Ebbini and Cain
1989). The location of the foci are in excellent agreement while the amplitudes of the
multiple foci varied by 20% in the 16 focus pattern and 50% in the 25 focus pattern. The
array gain (Ebbini and Cain 1989) (defined as the sum of the intensities of the focal
points divided by the total power for the array) tends to drop off significantly for patterns
which contain more than 8 foci for this array. Therefore, a series of multiple focus fields
with less than 8 foci were temporally switched to "fill in" a large volume in the transfocal
plane in the porcine experiments. The simulated and hydrophone scanned patterns are
found in Fig. 4-11. The focal amplitude vary less than 10%.
89
15
@0
0
-10-
E5
0
0
0,
0
0
0
c
0
.D 0
0
0
*0.*@
-5
-10-
o o~~C
00
-151
15
oo0*O@e0@,
o
000
Obe
C
0
e&o
0
ft
-
o
*00
E
C,)
0 .S,
0
0.
.b
-5
0
a
90
0
&
0
0
0
a
10
-151
-15 -10
-5
0
5
10
15 -15 -10
X-Axis (mm)
-5
0
5
10
15
X-A xis (mm)
Fig. 4-10: Simulated (left column) and hydrophone measured (right column) intensity field of 16 (top
row) and 25 (bottom row) focus patterns created in the focal plane of the array. The individual foci
were placed in a 5 mm spaced grid for the 16 foci and on a 4 mm grid for the 25 foci.
90
Simulated
(b)
5. (a)
-
Experimental
(g)
0@8
0
0.
E
.U)
0I
0
(h)
0
0
WV
we
0
0
-5-
e
m
U)0
5
0
-5
X-axis (mm)
5
0
-5
X-axis (mm)
-5
5
0
X-axis (mm)
-5
0
5
X-Axis (mm)
Fig. 4-11: Simulated (left) and corresponding hydrophone scanned (right) pressure magnitude
squared patterns in the focal plane used to create large focal volumes.
4.4 Discussion
4.4.1 Non-composite Piezoelectric Materials for "Dice and Fill" Arrays
The tests of LiNO and PZT indicate that, as expected, particular attention must be
taken to ensure proper width-to-thickness among piezoelectric elements. Elements whose
dimensions are close to the same size (factor of 3) will suffer significant drops in
efficiency.
However, elements with only one dimension that violates the proper
width-to-thickness ratio may be used if the drop in efficiency is still acceptable.
The
LiNO and PZT were both found to offer high efficiencies but the LiNO was about 10%
91
more efficient. The drawback of the LiNO is that it is a flat crystal and can not be ground
into curved pieces like the PZT. Therefore, PZT was chosen as the material for the
prototype arrays.
4.4.2 Kerfs Adhesives for "Dice and Fill" Arrays
Silicone (Loctite V-1022) can be used for arrays with large elements and epoxy
(Dow DER 732, DER 331, DEH 24) can be used for arrays with small elements (although
these arrays will fail over time due to epoxy cracking). A low coupling, robust adhesive
for diced arrays with small elements was not found for the dice and fill technique. The
most robust prototype arrays constructed as a part of this thesis used the Loctite Silicone
but it did not perform well over extensive power tests for arrays with a large number of
elements.
4.4.3 Electrical Connections
The choice of cable for a phased array is critical for its operation. The cable must
be non-magnetic to be put in an MRI, be flexible such that the array can be moved by a
positioning system, have low capacitance so that small array elements can be easily
matched to 50 fl, have low resistance to avoid power loss between the amplifier and
transducer, and have a small gauge to permit a large number of cables to enter the array
housing. The small gauge is particularly important for the 256 element array. While each
strand of the 34 AWG coaxial cable used for the 256 element array is able to deliver more
than sufficient power to the piezoelement, it also absorbs a significant amount of power,
thus degrading the electroacoustic efficiency of the overall transducer.
transmission line model can be used to predict the power loss.
92
The "L"
The loss could be
minimized by using shorter and larger gauge cable.
However, the larger cable bundle
and shorter length would make the transducer significantly more cumbersome and less
appropriate for clinical use in an MR scanner.
The designer must make a tradeoff
between electrical efficiency and ease of use. Since the power lost in the cable did not
significantly contribute to the heating of the piezoelectric element and the driving system
did not limit the electrical power to each element, it was decided to focus on making the
design of the 256 element array more "user friendly" instead of making the array more
efficient. There were little signs of electrical damage to the 256 element array or its cable
in the therapeutic experiments. As a final note on array construction: the length of cable
needed for the 256 element array is an impressive 1.8 km.
The best connection between a piezoelement and a cable was found to be a solder
joint.
It was able to withstand higher powers than the conductive epoxy or paints.
Likewise, the epoxy or paint electrodes tended to fail on the materials tested.
4.4.4 Composites from Material Systems, Inc.
The transducer performance from Material Systems varies greatly with the
composite fill ratio and the element width-to-thickness ratio.
For the materials whose
electroacoustic efficiency was low, the maximum output power which caused nonrecoverable damage to the piezoelement occurred at power levels below those needed for
therapeutic arrays. For example, the maximum power level was less then 3 W/cm 2 for
the 30% fill ratio. As a first order rule, the power should be kept at least to
V
of its
maximum. Therefore, of the four materials, only the 40% and 50% fill ratios can yield
output powers sufficient for therapeutic purposes. The high efficiencies for MS3 could
93
be the result of two aspects: 1) improved width-to-thickness ration or 2) an optimized fill
ratio.
The main failure mechanism which limited to the peak acoustical power from the
Material Systems composite materials was the silver epoxy electrode. When driven to
high powers, the PZT pillars cracked the epoxy coating and caused electrical arcs. The
pillars separated from the polymer either before or after this failure. Prior to complete
failure there was generally a decrease in element efficiency.
Tests of a prototype copper
electrode failed due to the lack of electrode adhesion to the polymer.
Coupling measurements taken for the 40% and 50% fill materials indicate that
inter-element coupling increases with increasing fill ratio. This is counter intuitive since
the inter-element spacing of the 50% fill material was greater than the 40% fill material.
It is possible that the larger pillar sizes of the 50% material may cross the kerf between
adjacent elements such that the electrode of one element is driving the pillar of an
adjacent element.
The coupling for either material is low enough that it does not
preclude the use of the composite materials in phased arrays.
In a comparison of the polymer matrix, the voided shore D80 PU matrix appeared
to yield slightly higher efficiencies in the 30% fill material than the other polymer and,
therefore, appears to be the least restrictive of the two polymers.
In conclusion, it was found that the MS3, 40% fill material supplied by Material
Systems could produce high enough acoustic powers to be used in a therapeutic array.
There were two large contraindications, however, against using this material.
First, it
contained a silver epoxy electrode which had been shown to fail over time. Second, the
94
company was unable to supply continuous pieces of transducers larger than 5 x 5 cm2 or
curved transducers.
4.4.5 Composites from Imasonic
The composite material supplied by Imasonic was found to be appropriate for
therapeutic phased arrays. It can generate high acoustic powers for extended periods of
time without damaging the array elements. The material is robust and efficient.
The nine element prototype array and the 256 element therapy array yielded
similar interelement coupling results.
Despite the measurement techniques used for
pulsed imaging arrays presented by the company which claim to offer better than 30 dB
isolation between adjacent elements, the acoustic efficiency measurements of adjacent
elements indicate that high power interelement coupling can cause non-negligible
changes in acoustic output power from the array. The piezocomposite polymer does not
isolate elements as well as silicone in a diced array. Nevertheless, although inter-element
coupling can decrease array output by up to 30%, the acoustic output power levels are
still sufficient for most therapeutic specifications.
A comparison between the acoustic measurements of prototype and therapy arrays
from Imasonic materials indicate that an acoustic quarter wave matching layer may not
offer significant advantages in a piezocomposite monofrequency therapeutic array. The
electroacoustic efficiency of the array without the matching layer is higher than the array
with the matching layer.
Since the arrays are air backed, the apparent loss in the
piezomaterial from multiple wave reflections is less than the transmission loss in the
matching layer. More importantly, the acoustic matching layer significantly increases the
95
undesirable interelement coupling in an array. The main advantage of using a matching
layer in the experiments of this thesis was to protect the ground electrode of the array.
After over 100 hours submersed in water, the copper/gold electrode of the 256
element therapeutic array showed little signs of wear.
4.4.6 Acoustic Fields from a Large Scale Array
Overall the 256 element array demonstrated that a large phased array can
accurately construct field patterns with up to 16 foci without causing excessive grating
lobes. There was good agreement between all of the simulated and hydrophone scanned
fields although amplitudes for multiple focus fields increasing varied as more foci were
produced. This could be due to the fact that the piezoelements were modeled as uniform
vibrators which they are not.
4.5 Conclusions
A technology to build a robust, large scale phased array was developed and tested.
The 1-3 piezocomposite material supplied by Imasonic offered a significant advantage to
diced PZT pseudocrystals in the construction of large scale therapeutic arrays.
The
electroacoustic efficiency and maximum power from the composite material was
sufficient for therapeutic applications. By containing a solid, uncut front, the composite
material did not suffer from undesirable leaking as did the cut PZT arrays while the
composite material still kept the geometry of the array intact despite its air backing.
Problems with the continuous ground plane were eliminated. Complex array patterns
could be created in the composite material without the need for expensive tooling. The
96
drawback of the material was its higher cost than regular PZT (the composite material
used in this research cost $4000 while comparable PZT bowls cost about $400).
97
5. TEMPORAL SWITCHING TO OPTIMIZE THERMAL DOSE
5.1 Introduction
Therapeutic phased array ultrasound transducers are advantageous in the treatment
of large tumors since they are capable of generating larger lesions than their non-phased
counterparts (Cain and Umemura 1986; Ebbini and Cain 1989; Fan and Hynynen 1995b;
Wan et al. 1996; Fan and Hynynen 1996a). The creation of these lesions is possible
through the use of multiple focus intensity patterns which distribute power and create
temperature elevations over regions much larger than those from a single spot focus. For
multiple focus therapy to be successful, however, care must be taken to reduce near field
heating (Hynynen et al. 1993; Damianou and Hynynen 1993; Fan and Hynynen 1995b)
and secondary temperature elevations (Ebbini and Cain 1989).
An earlier study
demonstrated that the rate limiting factors to necrose large tumors included both the
deposition of power in the tumor volume and the cooling period necessary between
sonications (Fan and Hynynen 1995b). If proper cooling times were not used, sequential
sonications could damage pre-focal tissue (Hynynen et al. 1993; Damianou and Hynynen
1993; Wan et al. 1996; Fan and Hynynen 1996a) or be blocked by thermally induced
cavitation between the transducer and the tumor volume (Hynynen 1991; Hynynen et al.
1993; Sibille et al. 1993; Malcolm and ter Haar 1996).
Thus although multiple focus
patterns have been shown to yield faster necrosis rates than single focus sonications, the
phased array treatments may still require extensive cooling times to produce a continuous
necrosed tissue volume (Wan et al. 1996; Fan and Hynynen 1996b).
98
To shorten the treatment duration and improve performance of phased array
ultrasound surgery, this study investigates a method to rapidly switch between multiple
focus patterns such that the thermal response of the array is more uniform, thus lowering
peak temperatures using less average acoustic output power. This technique is similar to
the temporal switching simulated for hyperthermia treatment (Umemura and Cain 1989;
Ebbini and Cain 1991 c), but differs in several key aspects. First, this switching technique
is designed for the short duration sonications required in ultrasound surgery. Therefore,
an important consideration for implementing this technique is the rate at which the
patterns are switched. Second, the primary goal of this research is to increase the tissue
treatment volume rate without producing undesirable therapeutic conditions.
These
conditions include excessively elevated temperatures in the tissue and high transducer
output power.
Third, while previous research has simulated and tested temporal
switching in a water bath, this research experimentally tests a temporal switching
technique in vivo for ultrasound surgery using temperature sensitive mapping provided by
non-invasive magnetic resonance imaging (MRI) (Kuroda et al. 1995).
5.2 Methods and Materials
5.2.1 Phased Array Design
A spherical sectioned array introduced by Ebbini and Cain (Ebbini and Cain
1991 a) and similar to the one described by Fan and Hynynen (Fan and Hynynen 1995b)
was designed and constructed for application in MRI guided surgery (Fig. 5-1). The array
had 16 elements and was sectioned from a single PZT-4 polycrystal and was matched to
50 Q near its resonance at 1.64 MHz using simple inductor-capacitor circuitry. The array
99
was powered by the phased array driving system with 8-bit phase resolution and self
leveling 0-60 W/channel power control.
R 8cm
A
Section AA
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
A
10 cm
Fig. 5-1: Spherical shaped square element array geometry.
5.2.2 Acoustic Measurements
The elements yielded acoustical efficiencies ranging from 70% to 80% at 2.5-3.5
W/cm 2 as measured by a radiation force technique (Stewart 1982). Beam plots in a water
bath were measured using a thermistor with a 0.25 mm silicone bead or by using a
0.5 mm hydrophone (NTR, Seattle, WA).
5.2.3 Numerical Simulation
Throughout this research, ultrasound fields were simulated using the RayleighSommerfeld integral over a set of geometrically superimposed point sources as described
by Zemanek (Zemanek 1971). The temperature elevations were calculated numerically
using the Pennes bioheat transfer equation (Pennes 1948) and the dose distributions were
calculated from a numerical integration of the Sapareto and Dewey model (Sapareto and
100
Dewey 1984) (see Appendix C for details). In all simulations, the spatial resolution was
at least 0.25 mm in the transverse axis and 0.50 mm in the longitudinal axis while the
temporal resolution was smaller than 0.02 s.
All calculations were performed on a
multiprocessor IBM PVS computer.
5.2.4 Optimization Routine
The goal of the optimization routine was to create a uniform dose over a region of
interest during a single sonication time period. To accomplish this, six patterns which
covered the possible region of sonication for the given array geometry were chosen as
inputs to the power optimization routine (Fig. 5-2).
5
5
(a)
-5
-5
0
5
X-Distance (mm)
(b)
5
-5
-5
5
5
(d)
(c)
0
0
-5,
-5
0
5
X-Distance (mm)
5
-5
-5
5
(e)
0
X-Distance (mm)
5
(f)
0
-5
-5
0
X-Distance (mm)
0
0
X-Distance (mm)
1
5
-51
-5
0
X-Distance (mm)
5
Fig. 5-2: Simulated fields for optimization generated using the mode scanning technique (McGough
et al. 1994).
101
The driving signals for these patterns were calculated using the mode scanning
technique which reduces near field heating by causing destructive interference along
different axial planes of the transducer (Cain and Umemura 1986; Umemura and Cain
1989; McGough et al. 1994). For example, pattern (b) of Fig. 5-2 was created by driving
all of the elements with equal amounts of power, but by varying their phases with
rotational symmetry. In this case, the center four elements (clockwise) are driven with
phases of {0,90', 180',270'} while the outer twelve elements (also clockwise) are driven
with 300 increments {0 ',30',...,2700,330'). This creates a destructive field pattern on the
longitudinal axis of the array since elements opposite from each other are driven out of
phase. Other patterns are created by varying the phase increment in the rotation or by
adjusting the element phases across any axis to create destructive interference (see Table
1 for the phase distributions of the patterns in Fig. 5-2). This technique has been shown
to decrease the required treatment time for a single sonication of a multiple focus pattern
as compared to other methods of pattern synthesis (Fan and Hynynen 1996b).
The
patterns were selected such that there was a distribution of foci throughout the proposed
treatment volume without extensive sidelobes (greater than 10% of desired foci) outside
of the three dimensional region of interest.
102
Table 5-1: The element phases (in degrees) used to create the fields patterns found in Fig. 5-2 (see
Fig. 5-1 for element numbering).
Element (Fig. 5-1)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
(a)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(b)
90
120
150
180
60
90
180
210
30
0
270
240
0
330
300
270
(c)
0
180
0
180
180
0
180
0
0
180
0
180
180
0
180
0
(d)
0
270
90
180
90
180
0
270
270
0
180
90
180
90
270
0
(e)
0
270
0
270
90
180
90
180
0
270
0
270
90
180
90
180
(f)
180
180
0
0
180
180
0
0
0
0
180
180
0
0
180
180
P1
0
0
0 0
0 0
P2
P3
Fig. 5-3: Switching technique diagram. On the left are the pressure plots for the six input patterns.
The power levels for these patterns (P1-P6) are determined through a gradient optimization and
rapid switching is then used to create an effective field as seen on the right.
103
Fig. 5-3 illustrates how the six patterns were used in the switching technique.
Basically, the driving signals rapidly changed between the six input fields to imitate a
continuous wave (CW) signal with more uniformly dispersed power in the region of
interest. The power levels for each individual field (P1-P6) were determined using a
direct-weighted gradient search routine (50 iterations) (Zahradnik 1971) to find the
"global" minimum of a mean-squared cost function comparing the simulated dose (Dsim)
to a uniform thermal dose (Dideal) over a region (V) slightly larger than the power
deposition patterns. (Eq. 1).
Cf
(Dsim (x,y,z) - Dideal(X, yZ))2
(Eq. 1)
In this case the region had a volume of 0.6 x 0.6 x 1.0 cm 3 . The ideal thermal dose of
5000 equivalent minutes at 43 0 C was chosen such that the threshold value for tissue
necrosis (240 equivalent minutes at 43 C) (Sapareto and Dewey 1984) was exceeded by
a factor of 20, thus reducing the chance of non-necrosed tissue within the region of
interest. The switched sonication duration was set at 10 seconds such that the simulations
matched the time necessary for the MRI to gather several temperature images during the
heating cycle (longer sonications were avoided to decrease perfusion dependence (Billard
et al. 1990; Dorr and Hynynen 1992)).
Several minutes passed between consecutive
sonications to avoid any effects of near field heating.
5.2.5 Switching Rate
Theoretical simulations were performed to find the effect of switching rate on the
effective dose distribution. To accomplish this task, the dose field was calculated for a
104
ten second sonication which consisted of various pattern dwell times ranging from 30-300
ms for six and for three input fields (the dwell time is defined as the time that each
pattern is used in one cycle of all of the input fields). The results of the simulation were
then compared to a simulation which had as its input an arithmetically averaged power
deposition pattern which was created by directly weighting the previously cycled input
fields. The goal of this study was to find the theoretical effect of the hardware switching
limitation speed on the generation of an effective continuous wave field.
5.2.6 Experimental Set Up Using MRI Thermometry
The phased array was placed in a submerged 3-dimensional positioning system
within a clinical 1.5 T Signa MRI (both GE Medical Systems, Milwaukee, WI) for
sonication in the thigh muscle of four New Zealand white rabbits in vivo (Hynynen et al.
1996). For each rabbit, the thigh was situated at the natural focus of the array and a series
of higher power sonications both with and without pattern switching were performed
while obtaining temperature sensitive images (Kuroda et al. 1995; Chung et al. 1996a).
These images used a fast spoiled gradient echo sequence with repetition time TR = 26.1
Ms, echo time TE = 12.8 ms, flip angle = 300, bandwidth = 7.2 kHz, resolution 128 x 256,
field of view FOV = 16 cm, and slice thickness = 3 mm. The time to obtain a single
temperature image was 3.3 seconds and some temperature elevation occurred during data
acquisition. The time sequence of images included a single image taken pre-sonication,
three images during the 10 second heating, and six images during the 20 second postsonication cooling period. After the temperature sensitive images were obtained, proton
density and T2-weighted images (fast spin echo sequence with TR = 2000 ms, TE = 17
105
and 68 ms, echo train length = 8, FOV = 16 cm, slice thickness = 3 mm) were taken to
demarcate the lesion areas and evaluate treatment execution.
5.3 Results
5.3.1 Simulation and Water Scanned Comparison of Array Fields
Phased array operation was initially tested by scanning the transducer in a water
bath. Fig. 5-4 compares a few of the theoretical and experimental scans. The four foci
pattern which forms a box in the focal plane demonstrates the off-axis spatial limits of
power deposition for this given array geometry, therefore outlining the maximum volume
which can be necrosed without physically moving the transducer. Discrepancies in the
focal width between the simulated and scanned results may be attributed to transducer
element misalignment and the relatively large thermistor cross section (0.25 mm). The
temporal switching operation of the hardware was confirmed by slowly switching
between the different multiple focus fields with the focal plane of the transducer located
at the surface of the water bath.
106
4
4
(b)
(a)
2
5
0
-2
-4
-4
_
-2
4
-41
-4
4
2
X-Distance (mm)
-2
0
4
2
0
X-Distance (mm)
4
-2
(c)
2
(d)
E2
0
0
-2
-4
-4
-2
0
2
X-Distance (mm)
4
-4,
-4
4
4
(e)
(f)
2
0
0
-2_
-2
-4
-4
4
0
2
X-Distance (mm)
-2
2
-2
0
X-Distance (mm)
4
-4
-4
0
2
-2
X-Distance (mm)
4
Fig. 5-4: Simulated and water scanned fields for 16 element phased array (a, c, e theoretical; b, d, f
experimental).
5.3.2
Optimization Results
The optimized power levels for the six patterns in Fig. 5-2 are listed in Table 5-2.
There are two main results of the optimization. First, the optimized power levels for each
field did not directly correspond to the number of foci of that given field. This is due to
an uneven overlap between the temperature and dose responses to the different power
deposition fields. Second, the optimal driving powers for three of the patterns {c, d, e}
were much larger than the other three patterns {a, b, f} such that the less substantial
patterns could be eliminated.
107
Table 5-2: Optimized switching power levels
Acoustic Pattern:
Fig. 5-2
(a)
(b)
(c)
(d)
(e)
(M
Optimized Power
(W
3
1
99
83
46
1
From a practical standpoint, reducing the number of switched fields is
advantageous since there is a limitation on how fast the hardware can toggle between
various phase and power patterns: 10 Hz was the limitation switching speed of the system
used in this research (the minimum sonication time per field was 100 ms). In order to
create an "effective" field which represents an arithmetic average of temporally switched
fields as if the field were CW, the switching frequency must become very fast as the
number of fields increases (see Fig. 5-5). For example, to avoid large errors between the
"effective" dose response produced by temporal switching and the theoretical dose
response from the average of the six input fields, the fields must be cycled faster than 600
ms (the cycle time is defined as the time to cycle through all of the input fields once).
This exceeds the switching rate of the hardware. The three fields, however, require a
cycle time below the hardware limitation.
108
3000
"0 2500
0
2000
cU
+3
1500
---
Fields MSE
6 Fields MSE
(D
0 1000
CM
2
500
00
0
50
100
200
150
250
300
Dwell Time for Each Pattern (ms)
Fig. 5-5: Normalized mean square error plots for dwell times comparing the temporally switched
"effective" field to the theoretical fields produced by an input field formed by arithmetically
averaging the three or six input fields respectively. The error was normalized by dividing it by the
optimal dose level.
The simulated thermal dose for the optimized switching pattern and a nonswitched pattern (pattern (c) of Fig. 5-2) of the same average power is found in Fig. 5-6.
It is clear from the simulation that the non-switched pattern yields much higher peaks and
lower dips in thermal dose within the region of interest.
One can also see that the
switching technique should create a contiguous lesion while the non-switched foci of the
same power would leave a center volume of non-necrosed tissue-an undesirable feature
when treating a large volume tumor. In order to create a completely necrosed region
using a simple four focus pattern, the power and post-sonication wait duration must be
increased more than 20% as listed in Table 5-3.
109
108
6
Six Fields Switched
4
-
10
10
10
Thermal
Necrosis
Threshold
-2
-15
-10
0
-5
5
10
15
Distance Accross Focus (mm)
Fig. 5-6: Optimization results of simulated dose across the focal axis. The four lines lines correspond
to the thermal dose delivered using a simple four focus pattern at 38 W (dashed), a four focus pattern
at 49 W (dotted), a switched pattern at 38 W using all 6 fields (solid), and a switched pattern at 38 W
using the three most significant fields (solid).
Table 5-3: Comparison of switched vs. non-switched fields to produce a continuous lesion of 240
equivalent minutes dose at 43 0 C.
Switched Pattern
Non-Switched Pattern (b)
38 W
W
49
Average Power
59s
71 s
Treatment Duration*
58 0 C
71 C
Peak Temperature
*Treatment duration includes the 10 second sonication and the amount of time for the
peak temperature to drop below 430 C.
Simulations with varying biological parameters tested the switching technique
under higher perfusion and tissue inhomogeneities. First, the simulated perfusion was
increased from 1 kg/m 3 s to 10 kg/m 3 s. For a static sonication of 49 W, the dose at its
center decreased from 265 equivalent minutes at 43C to 90 equivalent minutes,
110
indicating that even more power would be necessary to cause a continuous necrosis (see
The center dose of the 38 W switching technique also dropped but the
Fig. 5-7).
complete volume still exceeded the 240 minute threshold (also Fig. 5-7).
Second,
inhomogeneous tissue was modeled by creating a low absorption area (80% absorption as
compared to rest of tissue) with a cross section of 0.75 x 0.75 mm at the location of one
of the foci for the static and switched fields. The cross axial dose contours are graphed in
The switched contour exhibits a small decrease on the outer corner of the
Fig. 5-8.
necrosed region while the non-switched necrosis experiences a dip in thermal dose within
its outer borders.
10
106
.
Ct
Thermal
Necrosis
Threshold
10
102
0
100
10
10
-15
-10
5
0
5
10
15
Distance Accross Focus (mm)
Fig. 5-7: High perfusion simulation results (10 kg/Ms vs. 1 kg/M3s in Fig. 5-6). The 49 W static
sonication (dotted) dips below the necrosis threshold while the 38 W switched sonication (solid) still
creates a continuous necrosis region.
111
10.
10
10
10
5
5
U
0
0
-5
-5
-10
-10
-5
5
0
X-Axis (mm)
10
-10
-10
5
0
X-Axis (mm)
10
-5
5
0
X-Axis (mm)
10
10
10
5-
5
0
0
-10
-10
-5
-5
5
0
X-Axis (mm)
10
-10
-10
Inhomogeneous
Homogeneous
Fig. 5-8: Comparison of homogeneous tissue and inhomogeneous tissue with a small section of 80%
SAR at lower left foci. Contour lines correspond to dose of 15, 60, and 240 minutes for 38 W average
sonications.
5.3.3 MRI Experimental Results
Fig. 5-9 contains the temperature contour plots experimentally obtained using
MRI through the focal plane of a switched sequence sonication and a single four focus
sonication (pattern (c) of Fig. 5-2) of the same average power (76 W). One can see from
the two plots that the temperature distribution is more uniform for the switched pattern
than for the simple four focus pattern. Specifically, the center of the four focus pattern
112
~--~
~--~-
was thermally "filled in" by switching between multiple focus patterns instead of relying
on thermal conduction from the four distant outer foci. Both sonications yielded
continuous lesions as determined in post-sonication images (Fig. 5-10).
5
5
.-
E
E
0
0
C)
C)
CO)
0
-5
-5
d
-5
-5
5
0
Distance (mm)
5
0
Distance (mm)
Fig. 5-9: In vivo temperature contour images for (left) non-switched and (right) switched sonications
(contour lines at 10, 15, and 20 'C temperature increases) measured using MRI.
Fig. 5-10: Proton density weighted image of thermal necrosis caused by (a) single four focus pattern
and (b) switched focus pattern across axis.
113
To show that the switched fields could produce continuous lesions at lower
powers than a simple multiple focus pattern, a second set of 10 second sonications at 68
W were performed.
Fig. 5-11 was obtained after these lower power sonications. Note
that the switched pattern completely coagulated the treatment volume while the multiple
focus pattern left an unaffected region in its center.
This response may be better
explained by considering the temperature profiles across the sonication regions during
treatment (Fig. 5-12).
The switched pattern yielded a more uniform temperature
distribution and overall lower peak temperature than the single multiple focus pattern as
seen in simulations.
10
al
0"
0
Distance (cm)
10
Fig. 5-11: Proton density weighted image of lesions produced using 68 W average power: (a)
switched pattern lesion and (b) non-switched pattern lesion.
114
Switched Focus
Four Focus
30
30
(a)
25
25-
(b)
20
20CO,
CZ,
15 -
-
... . .15
Z 10-
10
E
5
5-
0-
0
-5
-10
0
Distance (mm)
-5
10
-10
0
Distance (mm)
10
Fig. 5-12: Temperature response across 68 W average sonications in vivo measured using MRI: (a)
four focus pattern and (b) switched focus pattern
5.4 Discussion
The main goal of this research is to use pre-treatment optimization to improve and
experimentally implement the use of temporal switching for a phased array. As a part of
this goal, this research presents the first quantitatively measured implementation of a
temporally switched ultrasound surgery treatment in vivo. Unlike previous switching
techniques, this research did not implement temporal switching only to increase the
effective focal volume of an array, but also to improve the treatment conditions such as
cooling duration, average power, and peak temperature.
The simulation and in vivo
results indicate that this goal can be accomplished: temporally switched fields can
115
decrease peak temperature and create regions of necrosis at lower average powers than
non-switched fields.
A second advantage of optimizing power among a set of deposition patterns is to
reveal those patterns which have the greatest effect on the thermal dose delivered to a
given volume. By indicating fields which have lesser effects, a smaller set of fields can
be implemented such that the required switching rates are not excessively fast. As is
expected, an "average" or effective field can be produced at lower switching rates when
fewer fields are used. A switching frequency of 10 Hz was fast enough to produce an
effective average field as described in (Ebbini and Cain 1991 c) for three fields but not for
six.
Therefore, given the hardware limitation, it is preferable to switch between the
smaller number of fields as long as the deletion of fields from the larger set does not
cause a drastic deviation from the ideal dose response of the complete set.
A third advantage to pre-treatment optimization is the indication of power levels
that are not necessarily dependent on the number of foci within the pattern. The precise
power levels of a set of patterns, however, is somewhat forgiving. It was found that a
variation of powers using the set {(c), (d), (e)} of Fig. 5-2 yielded similar thermal and
dose responses close to the global minimum. Not surprisingly, therefore, the choice of
deposition patterns appears to be the most critical feature when designing a treatmentthe gradient search having the ability to highlight those patterns of greatest utility and to
indicate the relative power levels for those patterns.
These levels can then be
proportionally scaled for in vivo treatments which may have varying or unknown tissue
absorption.
116
The temporal switching tested in this research could possibly be further improved
in two ways. First, the results presented in this study were all performed for a 10 second
sonication so that the theoretical and experimental results could be compared (the
limitation being the image access time of the MRI). The optimization would, therefore,
have to be repeated for different sonication times to correctly optimize the relative power
levels.
The sonication time was not optimized. Second, a variation of the switching
technique would optimize both the choice of field and its relative power for each time
segment in the switching period, imitating an adaptive power feedback control. In such a
case, the use of an ideal "arithmetic average" field as an "effective" field may not be
desirable since the power levels for each individual field would vary during the sonication
duration.
Both simulated and experimental results indicate that another reason to use
temporally switched fields is to decrease the dependency of tissue inhomogeneity and
biological parameters.
This is due to the fact that the switching technique distributes
power over a larger volume than the small number of concentrated foci from a static
pattern.
For example, in the simulation of the inhomogeneous tissue the static dose
response dropped more quickly between the four outer foci than did the dose response of
the switched dose pattern (see Fig. 5-8). Various samples of in vivo sonications in this
research also demonstrated that the dependency of tissue homogeneity may not be as
great when switching is used instead of static sonication (as an example, one of the foci in
the static sonication of Fig. 5-11 is greatly reduced).
It is hypothesized that a more
uniform power deposition decreases the chance that a temperature sink such as a blood
vessel will seriously effect the treatment when it spatially overlaps one of the static foci
117
as described in (Dorr and Hynynen 1992). Therefore, while the ability to spread power
distribution in ultrasound surgery over a large number of foci has previously been found
to be preferable from a treatment time consideration (Fan and Hynynen 1995b; Wan et al.
1996), it may also be preferable to ensure treatment operation.
Lastly, this research again showed that magnetic resonance imaging is a powerful
tool to detect temperature increases in vivo non-invasively. In particular, the temperature
images demonstrate the ability of MR imaging to detect temperature increases of
temporally switched fields, illustrating the feasibility of real time monitoring of a
dynamically changing sonication pattern. This has important implications toward real
time monitoring of ultrasound surgery.
118
6. A LARGE
SCALE
PHASED
ARRAY
SYSTEM
FOR MR
GUIDED
ULTRASOUND SURGERY IN THE LIVER
6.1 Introduction
This study investigated the use of a 256 element ultrasound phased array for the
coagulation of deep seated tissue in an in vivo porcine model (design, construction, and
acoustic measurements for the array are described in Chapter 4). It was the goal of these
experiments to verify the simulation results which have indicated that large tissue
volumes that could be coagulated in a single sonication using a large scale array (Wan et
al. 1996; Fan and Hynynen 1996a) and to demonstrate the feasibility of ultrasound liver
surgery using a phased array. The array was used in a series of in vivo and ex vivo / in
situ porcine liver experiments.
6.2 Materials and Methods
6.2.1 Numerical Simulations
The ultrasound fields of the 256 element array were simulated using the RayleighSommerfeld integral over a set of geometrically superimposed point sources as described
by Zemanek (Zemanek 1971). The acoustic vibration on the surface was modeled as
uniform.
bioheat
The temperature elevations were calculated numerically using the Pennes
transfer
equation
(tissue
constants:
perfusion = 1 kg/m 3/s,
thermal
conductivity = 0.48 W/m/C, arterial blood temperature = 33.50 C, specific heat of tissue
and blood = 3770 J/kg/0 C, density = 998 kg/m 3 , and ultrasound attenuation = 0.041
119
Np/cm/MHz (Duck and Perkins 1988; Moros and Hynynen 1992)) and the dose
distributions were calculated from a numerical integration of the Sapareto and Dewey
model (Pennes 1948; Bowman et al. 1975; Bowman 1981; Bowman 1982; Sapareto and
Dewey 1984) (see Appendix C for more details).
In all simulations, the cross-axial spatial resolution was 0.5 mm, the along-axis
spatial resolution was 1.0 mm, the temporal resolution was 0.02 s, and the region of
calculation extended from 3 cm to 13 cm from the array and ±4 cm from the axis of the
array. The large simulation volume was necessary to avoid excessive simulated cooling
from the region's boundaries. To compare the simulated temperature elevations to the
MR temperature images, the simulation results were averaged using a uniform spatial
filter of the MR voxel size. The phase distribution for the array was calculated using the
pseudoinverse technique (Ebbini and Cain 1989).
magnitudes.
The elements had uniform power
All calculations were performed on a dual 300 MHz Pentium II
PC
(Micron, Boise, ID).
6.2.2 Porcine Model
A porcine model was chosen for the experiments to geometrically approximate
size of the human anatomy. Six pigs ranging from 30-40 kg were anesthetized using an
intramuscular injection (ketamine, 15 mg/kg; xylazine, 2.2 mg/kg; atropine, 0.05 mg/kg)
followed by an intravenous drip to the dorsal auricular vein (ketamine, lmg/ml; xylazine,
lmg/ml; guaifenesin, 50 mg/ml; 5% dextrose; rate of 2.2 ml/kg/hr). The thigh, abdomen,
and/or back of the pig was shaved and cleaned to create a clear acoustic window for
ultrasound transmission. The pig was intubated with a 7 mm endotracheal tube to ensure
120
a patent airway but no respiratory assist was given.
A naso-gastic tube was used to
suction out any air or gastric fluids which extended the stomach such that it blocked the
available acoustic window. The animal protocol was approved by the Harvard Medical
Area Standing Committee on Animals according to NIH and Harvard Medical School
guidelines.
In addition to the in vivo animals described above, two pigs were used in ex vivo /
in situ sonications of the liver. These experiments were begun less than thirty minutes
after the pig expired and were performed across the abdominal wall of the intact animal
corpse.
Following the experimental protocol the animal was sacrificed and dissected.
Gross measurements of the ultrasonic lesions were made and histological samples were
taken and stained with hematoxylin and eosin for analysis.
6.2.3 MR Experimental Set Up
The in vivo or ex vivo animals were placed in the bore of a clinical 1.5 Tesla Signa
MR imager (GE Medical Systems, Milwaukee, WI).
Fig. 6-1 is a diagram of the
experimental set up in the magnet. The animal is coupled to the array through a water
bath suspended above the array. A 12.5 cm diameter MR surface coil (GE Medical
Systems, Milwaukee, WI) was used to improve the imaging signal. Prior to sonication,
fast
spin-echo T2-weighted
images (TE/TR = 72/2000ms,
echo train length = 8,
FOV = 20 cm, thickness = 3 mm, matrix size = 256x256, NEX = 2, bandwidth = 16kHz)
or SPGR images (SPoiled Gradient Recalled acquisition in steady state; slice
thickness = 3 mm, FOV = 20 cm, TE/TR = 7. 1/100 ins, echo train length = 1, NEX = 1 or
121
2, flip angle = 450, bandwidth = 3.1 kHz) were used to locate the array in relation to the
animal. Temperature sensitive images were taken during a non-destructive low power
sonication to locate the array focus and determine the target tissue (proton resonant
frequency shift constant = 0.00909 ppm/0 C; slice thickness = 3 mm, FOV = 20 cm, image
acquisition time = 6.66 s, TE/TR = 24.3/49 ms, NEX = 1, flip angle = 300, echo train
length = 1, bandwidth = 3.1 kHz). Temperature images were then acquired during the
sonication time and during the cooling time (typical total imaging time of 150 seconds for
23 temperature images including three images from the 20 second sonication time).
These temperature images were used to calculate the predicted thermal dose and tissue
necrosis using the Sapareto and Dewey model (Sapareto and Dewey 1984).
Post
treatment images (T2-weighted) were taken to evaluate the tissue response and measure
tissue damage.
Amplifier System
* 512 channels
* 0-60 watts/channel
* 20 phase shifting
Porcine Model
MRase
Curfl
Water
.
50 Q LC Matching
2-Axis Positioning
System
Signa 1.5 T
Fig. 6-1: MR experimental design.
122
6.2.4 Ultrasound Surgery Experiments
The experiments were divided into four classes: in vivo thigh muscle (n=5), in
vivo liver (n=4), in vivo kidney (n=2), and ex vivo / in situ liver (n=2). All of the results
presented were produced in sonication periods of 20 seconds unless otherwise noted. The
in vivo thigh muscle and ex vivo / in situ experiments were used to test the array under
respiratory motion free conditions.
The ex vivo series of sonications were performed in
the liver of an expired pig to investigate the ability of magnetic resonance imaging to
detect heating in the acoustic obstacles and to evaluate the available acoustic window.
The liver sonications were performed using an acoustic window inferior to the sternum
into the right and left inferior lobes of the liver, typically medial to the gall bladder. The
array was positioned by using the SPGR images to detect bone and cartilage bodies. The
kidney sonications were performed inferior to the rib cage and lateral to the vertebral
bodies. The kidneys were located using T2 and TI weighted MR images and the focal
region of the transducer was positioned at the deepest location in kidney possible.
Intentional sonications near bone or cartilage are noted in the results.
A series of multiple focus fields were temporally switched to "fill in" a large
volume in the transfocal plane in the porcine experiments. The simulated patterns are
found in Fig. 4-11.
The phase distribution of the array was determined by using the
pseudoinverse technique with a phase rotation of the desired focal pressures about the
array axis (Ebbini and Cain 1989).
The first protocol used patterns (a)-(c) to fill in a
volume with 0.5 x 0.5 cm 2 cross section (9 effective foci) while the second protocol used
patterns (b)-(f) to fill in a volume with a 1.0 x 1.0 cm 2 cross section (24 effective foci).
The spacing between foci in the effective grid of foci was chose to be 2.5 mm such that
123
the thermal dose between foci would exceed the thermal dose threshold of coagulation
with a peak focal temperautre less than 650 C. Table 6-1 and Table 6-2 show the relative
power levels used for the respective fields. The fields were optimized such that the peak
intensity of each pattern was the same. Dose optimization was not employed due to its
excessive computational needs (the IBM PVS was no longer available for use).
The
center focus of the large focal volume was not applied as the thermal conduction from the
other patterns would heat the center of the focal volume. The patterns were switched at a
rate of 18 Hz.
Table 6-1: Relative powers used in a temporally switched field with 9 effective foci in a 0.5 x 0.5 cm 2
area in the focal plane.
Relative Power
1.00
4.46
4.46
Pattern (from Fig. 4-11)
(a)
(b)
(c)
Table 6-2: Relative powers used in a temporally switched field with 24 effective foci in a 1.0 x 1.0 cm
area in the focal plane.
2
Relative Power
1.00
1.00
1.11
2.60
1.48
Pattern (from Fig. 4-11)
(b)
(c)
(d)
(e)
(f)
6.3 Results
6.3.1 In Vivo Thigh Muscle Experiments
The array was used to electronically shift the sonication focus in the porcine thigh
muscle while recording temperature sensitive MR images.
Fig. 6-2 contains the
experimental temperature images and simulated temperature fields for a single focus
124
which is electronically shifted along the axis of the array.
The power necessary to
experimentally generate the in vivo temperature elevations was 1.8 to 3.5 times higher
than the power predicted through simulations (see figure caption for power levels). Fig.
6-3 shows the T2-weighted images of the resulting lesion from the axial sonication along
with the simulated and experimentally predicted lesion size (measured as 4.5 x 0.7 xO.4
cm 3 ).
140
140.
140
140
r20
120
120
120
'100
100
100
100
80
80
80-
80
60
-20
0
20
X-Axis (mm)
60
-20
0
20
X-Axis (mm)
60
601
(a)
-20
0
20
X-Axis (mm)
(b)
(c)
-20
0
20
X-Axis (mm)
(d)
Fig. 6-2: Experimental temperature images (top) and simulated temperature images (bottom) of onaxis electronic shifting of a single focus. The sequential sonications from left to right were placed at a
distance of 10, 9, 12, and 11 from the transducer which was placed 3 cm from the skin of the porcine
thigh. The peak temperature elevation were measured to be 370 C, 270 C, 27* C, 380 C from left to
right for input powers of 61 W, 61 W, 146 W, and 85 W respectively. The simulated fields were
driven such that their peak temperature matched the experimental results. The simulated powers
were 34 W, 23 W, 41 W, and 38 W from left to right.
125
..
.......
Fig. 6-3: Lesion produced from the axial shifted sonications of Fig. 6-2. The middle images overlays
the simulated predicted lesion size corresponding to a thermal dose of 240 equivalent minutes at
430 C. The right image shows the predicted lesion using the temperature images obtained during
sonication and cooling applied to the Sapareto and Dewey model (same thermal dose level).
Fig. 6-4 and Fig. 6-5 are temperature images which demonstrate the ability to shift
a single focus both across the axis in the focal plane.
Fig. 6-6 shows the resulting
temperature lesions generated in the thigh. There was some interference caused by the
muscle interface in the placement of the off axis foci. Individual foci of multiple focus
patterns were detectable using temperature sensitive images in the thigh muscle. Fig. 6-7
shows two patterns of four foci spaced at 5 mm and 20 mm.
126
Fig. 6-4: Temperature images of off axis focusing of a single focus in porcine thigh. The top left
image measured a 200 C temperature rise (subnecrosis since the porcine temperature was measured
as 33.50 C) at the natural focus of the array (used to localize the focus in the tissue). The top right
and bottom images correspond to off axis electrical focusing 7 mm away from the natural focus (the
white "x" corresponds to the location of the geometric array focus). The peak temperature for the
off axis foci ranged from 40-450 C.
127
Fig. 6-5: Axial temperature image of a focus shifted off axis. The "x" corresponds to the location of
the natural focus. This location would be above and to the right of the natural focus in the previous
figure. The peak temperature elevation was 360 C.
Fig. 6-6: Lesions produced from off axis electrical focusing (see previous two figures).
128
Fig. 6-7: Multiple focus patterns in focal plane of porcine thigh. Each pattern contains four foci at
the corners of a square. The spacing between foci is 5 mm (right) and 20 mm (left). The peak
temperature elevations were 6 0-7* C.
2
A sample of the temperature response of the switched 0.5 x 0.5 cm focal volume
is found in Fig. 6-8. The figures show the temporal and spatial profiles of the cross axial
heating produced in the thigh for an average sonication of 107 W of acoustic power (see
Table 6-1 for relative powers). Fig. 6-9 shows the post sonication images of the lesion.
Fig. 6-10 shows a photograph of the lesion. The lesion cross section measured 0.7 x 1.7
cm2. Images of a second sonication using the same sequence (but only 53 W) were taken
in the axial plane of the array. The temperature response and lesion images are found in
Fig. 6-11 and Fig. 6-12.
129
60
.
40
20
0
0
S60
60
~40
40
20
60 80
40
Time (sec)
100
5 0.
& 20
s 20
0
-23.4
0
-21.4
23.4
11.7
0
-11.7
Accross Axis (X) (mm)
-11.
23.4
11.7
0
Across Axis (Y) (mm)
2
Fig. 6-8: Across array axis temperature response in vivo thigh muscle for a 0.5 x 0.5 cm focal
pattern. The images in the upper left correspond to the first four temperature images taken during
sonication. The lower right image of the upper left set contains the calculated thermal dose contours
(240 and 2000 equivalent minutes at 430 C). The other plots correspond to the peak temperature
profile over time and the focal spatial response at end sonication.
Fig. 6-9: Images of a 0.5 x 0.5 cm2 cross section lesion formed in porcine thigh. The upper left image
is the full temperature image of the thigh at end sonication. The upper right image is a T2-weighted
image in the same plane as the temperature image with overlying contour of estimated thermal dose
(240 eq. min. at 43' C). The bottom right image is that same image without the contours (notice the
existence of a similar lesion 2 cm to the right). The lower left image is a contrast enhanced T1
weighted image of the lesion along the axis of the array.
130
Fig. 6-10: Photograph of lesion in porcine thigh caused by the 0.5 x 0.5 cm 2 acoustic pattern.
40
*
S
30
I.
S
0
20
10
0
20
40 60 80
Time (sec)
100
60
4G
'.1
A(
4
'U
2
=1
E
0
-23
0
-23.4 -11.7
0
11.7 23.4
Along Axis (mm)
-11.7
0
11.7 23.4
Cross Axis (mm)
Fig. 6-11: Along array axis temperature response in vivo thigh muscle for a 0.5 x 0.5 cm 2 focal
pattern. The images in the upper left correspond to the first four temperature images taken during
sonication. The lower right image of the upper left set contains the calculated thermal dose contours
(240 and 2000 equivalent minutes at 430 C). The dip (arrow) in the lower right temperature plot is
an imaging artifact (tissue interface).
131
Fig. 6-12: T2-weighted images of lesion formed in sonication from Fig. 6-11.
same with the exception of the overlaying thermal dose contours.
The images are the
Fig. 6-13 shows a sample of the temperature sequence of images and the
corresponding temperature response in the thigh when the effective 1.0 x 1.0 cm 2 cross
section focal pattern was used in a 20 second sonication (average acoustic power from the
entire array was 345 W-see Table 6-2 for relative powers). Fig. 6-14 and Fig. 6-15
contain the spatial and temporal response of the sonication, respectively. The average
power used in the simulations was the same as the experimental power.
Fig. 6-16
contains the T2-weighted images post sonication along with overlaying thermal dose
contours predicted using the simulation data and the set of experimentally measured
temperature images. The simulated lesion was slightly smaller than the estimated lesion
size. Both the MR images and post mortem dissection indicated a lesion size of 3.2 cm x
1.3 cm x 1.3 cm (5.4 cm 3) extending from 5.5 to 8.7 cm underneath the skin interface.
Similar lesions were formed as close as 3 cm from the skin surface. Fig. 6-17 contains a
T2-weighted image across three of the large volume sonication lesions.
contains a summary of the 1.0 x 1.0 cm 2 lesions.
132
Table 6-3
20.0 s
13.3 s
60.Os
100.Os
140.0 s
Fig. 6-13: Series of temperature images along the array axis during a 20 second sonication period
(top) and during the first two minutes of the cooling period (bottom). The array is still underneath
the porcine thigh although the orientation of the MR images places the array on the left as sonicating
from left to right.
0
) 30
j:
0
R 20
-
*Q
30
201
'.
Ib*
0
30
50
A
E 10
1,4 10
70
90
110
Along Axis (mm)
130
,
0000.
-30-20-10 0 10 20 30
Across Axis (mm)
Fig. 6-14: End sonication spatial temperature response of the 1.0 x 1.0 cm 2 cross section focus. The
solid line corresponds to a the simulated temperature response and the circles correspond to the MR
measured temperature elevations. The arrow indicated the location of muscle interface tissue where
the MR signal does not correlate with temperature.
133
co 35 30 25
20
15
10
5
0
-5
0
20
40
60 80 100 120 140
Time (s)
10
V
8
6
I-
- - - - - - - - - - - - - - - - - - - - -
4
2
0
20
40
60 80
Time (s)
100 120 140
Fig. 6-15: Temperature elevations at end sonication in the focus (top) and in the prefocal tissue
(bottom). The time plots on the right correspond to the simulated (dashed) and experimental (solid)
average temperature in the boxes overlaying the respective images on the right.
134
Fig. 6-16: T2-weighted images of the 1.0 x 1.0 cm2 lesion in the thigh. Image (a) is along the axis of
the array and image (b) is in the focal plane. Images (c) and (d) are the along axis image with
overlaying thermal dose contours corresponding to 240 and 2000 equivalent minutes at 430 C using
the experimentally measured temperature images (c) and simulated temperature field (d).
Fig. 6-17: T2-weighted image of the cross section of three large volume sonications (arrows). The
lesions used the 1.0 x 1.0 cm 2 pattern with an average power of 260 W. The average width was
13 mm.
135
Table 6-3: Summary of large focal lesions with the exception of the ten overlapping large volume
sonications.
1
2
3
4
5
6
7
8
9
10
11
Average Acoustic Power
(W)
172 (40 s)
172 (20 s)
259 (20 s)
259 (20 s)
259(20s)
302 (20 s)
259(20s)
259 (20 s)
302 (20 s)
259 (20 s)
259 (20 s)
Lesion depth
(cm from skin)
2
12
13
345 (20 s)
345 (20 s)
5.5
5.5
#
-
4.2
5.1
4.8
5.2
5.1
4.9
5.5
-
-
MR measured
dimensions (cm 3)
3.2x1.2x1.2
no lesion
round 1.1
3.1x1.2x1.2
2.9x1.lxl.1
3.1xl.4x1.2
3.1xl.2x1.2
2.9x1.Ox1.0
2.5x1.lx1.0
no lesion
lesion at prefocal muscle
interface
3.4x1.2x1.2
2.7x0.9x0.9
Volume
(cm 3)
4.6
0
0.69
4.4
3.5
5.2
4.4
2.9
2.7
0
0
Peak Temperature
Elevation ("C)
32
20
22
24.5
20
26.1
33.1
27.6
20
17
10
4.9
2.1
35
24.6
Some of the large focal volume sonications were altered by tissue interfaces. Fig.
6-18 shows the results of three sonications (in rows) made near the muscle interface of a
1.0 x 1.0 cm 2 focal volume protocol with greater than 30 minutes between successive
sonications. From left to right, the images show the end sonication temperature image, a
T2-weighted image with estimated dose contours of 240 and 2000 equivalent minutes at
430 C, and a T2-weighted image of the produced lesion. The average temperature in the
muscle just under the skin surface was measured from the MR images to be 2.2, 2.5, and
2.3 0C, respectively, verifying that similar power levels were delivered through the skin
for each of the sonications. The peak temperature in the focal zone was measured as
33.1, 15.2, and 14.7 0C (from top to bottom) indicating that the ability to focus at the set
depth was compromised for the later two sonications. The first row of images show a
lesion whose shape was skewed near the interface. The second row of images show a
lesion that is barely visible and not well formed. The third row of images show a nondistinct lesion beyond the interface and a definite lesion in front of the interface.
136
I
Fig. 6-18: Three large focal region sonications close to a muscle interface. The focal pattern covered
a 1.0 x 1.0 cm2 area in the focal plane of the array. The average acoustic power for each of the
sonications was 259 W. The images on the left are the temperature images at peak temperature, the
center images are the T2-weighted images of the lesion with overlying thermal dose contours, and the
right images are the T2-weighted lesion images without the contours.
A series of 10 adjacent large focal volumes (1.0 x 1.0 cm 2 ) were performed in the
thigh of one pig. The average acoustic power was 345 W. After each sonication, the
array was moved 7.5 mm. Each sonication lasted 20 seconds and there was an average of
9 minutes between consecutive sonications. Fig. 6-19 shows the temperature response
137
and lesion formed for the first two lesions. Fig. 6-20 is a T2-weighted image of the
complete lesion. It measured 3.8 x 2.2 x 3.0 cm 3 (25 cm 3 ). On gross examination the
lesion was the same size.
Fig. 6-19: Images of the first (left) and second (right) sonication of the ten overlapping sonications.
The top images are the MR temperature images at end sonication. The middle images contain
the
resulting T2-weighted image with overlapping thermal dose contours of 240 equivalent minutes
at
430 C estimated using the MR temperature images. The bottom images contains the
resulting T2weighted images with and without the cumulative thermal dose contour.
138
I
Fig. 6-20: T2-weighted images of lesion formed by 2 rows of 5 large volume sonications. The images
are taken in three planes of the rectangular lesion shape. According to the MR images, the lesion
3
measures 3.8 x 2.2 x 3.0 cm 3 (25 cm ).
Histological sample of the muscle lesions presented as fragmented cells with
disorganized, non-fibrous, non-striated cytoplasm (see Fig. 6-21). The lesion nuclei were
pyknotic. There were some cells within the lesions that appeared to have less damage
than the surrounding cells. The results were consistent with other studies (Hynynen et al.
1993; Hynynen et al. 1994).
Fig. 6-21: Hematoxylin and eosin stained muscle tissue. The tissue on the left is healthy muscle and
the tissue on the right is from a thermal lesion.
139
6.3.2 In Vivo Kidney Experiments
One or two sonications were performed in each of the kidneys of two pigs (total of
seven lesions). Due to the small thickness of the kidney cortex, the focal region was
limited to a 0.5 x 0.5 cm2 cross section in the focal plane (average power of 210 W). Fig.
6-22 contains the MR images of the end sonication temperature and thermal lesion of one
of the sonications. A photograph of that lesion is found in Fig. 6-23. Another of the
kidney sonications was partially blocked by the intestines and spinal vertebrae. Fig. 6-24
shows the temperature response and lesion formation from this sonication. Table 6-4
contains a list of the kidney lesions.
Histological evaluations of the lesions presented cells with pyknotic nuclei and
extensive hemorrhagic congestion into the tubules and glomeruli (see Fig. 6-25).
The
cells contained large vacuoles and the tubules appeared enlarged. The renal findings were
consistent with those found for a single focus transducer (Susani et al. 1993; Hynynen et
al. 1995).
Fig. 6-22: MR images of a kidney sonication. The image on the left is the temperature response of a
0.5 x 0.5 cm 2 cross section focus (the peak measured temperature elevation was 240 C). The resulting
lesion is viewed in the T2-weighted image in the middle with the estimated thermal dose contours at
240 equivalent minutes at 430 C on the right.
140
Fig. 6-23: Photograph of kidney lesion produced from sonication viewed in Fig. 6-22.
Fig. 6-24: Kidney sonication. The temperature image on the left indicates excessive heating in the
lower left corner of the kidney next to the intestines and on the right side next to the vertebral body.
The T2-weighted image in the middle shows a small lesion in the kidney and tissue damage
surrounding the vertebrae. The estimated thermal dose contours (240 eq. min. at 430 C) are found on
the right.
141
Fig. 6-25: Microscopic slide of kidney glomeruli stained with hematoxylin and eosin. The tissue on
the left is healthy kidney tissue and the tissue on the right is of the hemorrhagic thermal lesion.
Table 6-4: Kidney lesions.
Organ
ID
P7S61
P7S78
P7S81
P8S49
P8S59
P8S63
P8S68
Focal Pattern (cm x
cm)
0.5 x 0.5
0.5 x 0.5
0.5 x 0.5
0.5 x 0.5
0.5 x 0.5
0.5 x 0.5
0.5 x 0.5
Power
(W)
212
212
212
212
212
212
212
Temp.
26
28
30
25
25
23
40
MR
Lesion
(cm x cm)
1.0 x 0.7
1.1 x 0.7
0.8 x 0.6
0.9 x 0.5
1.1 xO.7
1.0 x 0.5
1.2 x 0.5
Autopsy Lesion
(cm x cm x cm)
1.0 x 0.6
1.2 x 0.7
Not found
Not found
Not found
1.0 x 0.5
1.0 x 0.5
Comment
Vertebral blocking
On edge of air filled
intestine
6.3.3 Ex Vivo/ In Situ Liver Sonication
Fig. 6-26 shows the heating pattern of the large volume focus (1.0 x 1.0 cm 2 cross
sections with an average acoustic power of 302 W) in ex vivo / in situ liver. The peak
temperature elevation was 260 C. The T2-weighted image does not show a clearly
outlined lesion. The transducer was repositioned and the focal region cross section was
decreased to 0.5 x 0.5 cm 2 (average power of 212 W). Images of the sonication and
resulting lesion are found in Fig. 6-27. In this case, the much higher temperature left a
lesion clearly visible in the MR image.
142
"Web-
30-
30
m
0
$i'
0
20
10
t
2010
-AJ
0
0
11.7 23.4
0
-23.4 -11.7
Cross axis (mm)
11.7 23.4
0
-23.4 -11.7
Along axis (mm)
Fig. 6-26: Large focal region sonication in ex vivo / in situ liver. The lower plots indicate the peak
temperature rise across and along the axis of the array at the focal center. The top, right images are
of the lesion imaged post sonication with and without thermal dose contours (damage not seen
clearly). To reduce noise in the dose calculation, a temporal filter was used to average pixel
temperatures when temporally adjacent measurements were greater than 300 C different.
(right) in ex vivo
Fig. 6-27: MR images of the end sonication temperature (left) and generated lesion
2
MR measured
The
.
cm
0.5
x
0.5
of
section
cross
a
square
had
region
focal
liver. The attempted
at an average
C
530
was
elevation
temperature
peak
The
long.
cm
1.2
and
wide
lesion was 0.5 cm
430 C) are on
at
min.
eq.
(240
contour
dose
thermal
lesion
estimated
The
W.
212
of
power
acoustic
the right image.
143
The ex vivo porcine models were used to investigate the ability of MR to evaluate
available acoustic windows. An SPGR sequence was used to find the costal and chondral
ribs as shown in Fig. 6-28. The 54 second scan time was decreased for in vivo images by
changing the number of MR excitations from 2 to 1 and by decreasing the phase encoding
resolution from 256 to 128. This decreased the imaging time to 13.5 seconds-a more
appropriate breathe hold time to evaluate the acoustic window into a patient.
Fig. 6-28: SPGR image of ribs (left) and temperature image (right) of rib heating.
Without localizing both the boney and cartilaginous ribs, undesirable heating can
be caused. This heating may also be detected by MR. Fig. 6-28 shows a sonication that
passed through a rib causing a prefocal heating. Fig. 6-29 shows the sonication heating
response in the cross axial plane of ribs at the costal-chondral junction inferior and lateral
to the sternum. The average temperature response at several points in the tissue is plotted
in Fig. 6-30. The lower two plots of the temperature in bone and cartilage are probably
quantitatively inaccurate but may be proportional to the accurate temperature since the
proton resonant frequency constant was assumed to be that of soft tissue for all the plots.
144
The soft tissue of the intercostal space (top image and plot) is probably the most accurate
temperature measurement. Note that the intercostal space continues to heat up after the
20 second sonication in response to heat being conducted from the adjacent ribs.
Fig. 6-29: Rib heating during sonication. On the left is a temperature image taken in the rib plane
between the transducer and the focal volume during a large volume sonication. On the right is an
SPGR image of the costal and chondral ribs.
U12
8
4
0
12
8
ci4
0
U
50
40
30
cL 20
S
410
0.
01
40
80
Time (s)
120
Fig. 6-30: Average temperature elevations in the rib plane for a high power sonication. The
temperature vs. time plots on the right correspond to the average temperature of the pixels in the
small boxes in the images on the left. The top box is in the intercostal space, the middle box is over
chondral ribs, and the bottom box is on the costal ribs. All of the images assumed the same proton
resonant frequency shift constant.
145
6.3.4 In Vivo Liver Experiments
The array was targeted in the porcine liver using T2-weighted and SPGR MR
images. Several acoustic fields were used to sonicate the porcine liver. Fig. 6-31 shows a
2
sample of the spatial temperature response at peak temperature for the 0.5 x 0.5 cm
pattern at an average power of 212 acoustic watts. A photograph of the lesion is found in
Fig. 6-32.
Fig. 6-33 shows the temperature response of the liver when a series of
0.5 x 0.5 cm2 cross section focal regions were placed at 11, 10, 9, and 9.5 cm from the
3
base of the array. The resulting lesion was measured to be 4.0 x 0.6 x 0.8 cm both in
MR images and in gross histology. The 1.0 x 1.0 cm 2 cross section focal pattern was
applied with an average power of 345 W. The peak temperature rise in the tissue was 280
C. The MR images and a photograph of the lesion are found in Fig. 6-34 and Fig. 6-35.
On the MR image the lesion cross section measured 0.9 x 1.3 cm 2 . The lesion measured
by gross histology was 2.6 x 1.1 cm 2 . A list of the multiple focus liver sonications is
found in Table 6-5. Note that not all of the lesions appeared in the T2-weighted images
but the majority of high power sonication did produce lesions that were identifiable using
the MR images.
146
Fig. 6-31: MR images of a 0.5x0.5 cm 2 cross section focus in vivo liver. The top image is the
temperature response at end sonication (peak temperature elevation 340 C at 212 W average acoustic
power). The middle image is a T2-weighted image of the thermal lesion with overlying dose contours
of 240 and 2000 equivalent minutes at 43* C. The bottom image is the T2-weighted image of the
lesion without the thermal dose contours. The MR measured cross section of the lesion was 1.2 x 0.6
2
cm2.
147
5 '6'7
le
Fig. 6-32: Photograph of liver lesion formed in vivo. The lesion was formed from the sonication
2
described in Fig. 6-31. It measured 1.5 x 0.8 cm .
2
Fig. 6-33: The top row contains temperature images of 0.5 x 0.5 cm foci located (from left to right) at
a distance of 11, 10, and 9 cm from the array. The peak temperature for the sonications was 390, 430,
and 34' C (left to right). The lower row contains the T2-weighted image of the lesion following the
sonications with the cumulative thermal dose contours (240 eq. min. at 43* C). The end lesion
measured using MR and gross histology was 0.6-0.8 cm wide and 4.0 cm long.
148
Fig. 6-34: End sonication temperature image and Ti-weighted contrast enhanced image of the lesion
for a 1.0 x 1.0 cm 2 sonication area in vivo liver. The peak temperature elevation was 280 C. The
image on the right contains the estimated dose contour (240 equivalent minutes at 430 C).
Fig. 6-35: Photograph of lesion formed from a 1.0 x 1.0 cm 2 cross section focus in the focal plane of
the array Fig. 6-34. The peak temperature was 280 C. The lesion is well defined but the center is not
homogeneous as the smaller lesion of Fig. 6-32.
149
Table 6-5:Table of in vivo liver multiple focus sonications and lesions.
Organ
ID
Focal Pattern (cm x
cm)
Power
(W)
115
Temp.
Ele.("C)
27
Lesion
MR
(cm x cm)
Not visible
Autopsy Lesion
(cm x cm x cm)
2.5 x 0.7 x 0.7
Comment
large
Overlaps
4 foci at corners of
vessel
0.5 x 0.5 cm 2 box
30 s sonication
None
15
173
1.0 x 1.0
P5S63
30 s sonication
None
15
173
1.0 x 1.0
P5S64
None
Not visible
107
0.5 x 0.5
P5S66
1.1 x 0.6 x 0.8
Not visible
33
160
0.5 x 0.5
P5S67
See Note*
1.4 x 0.5
Not visible
212
0.5 x 0.5
P7S4
See Note*
Not visible
>40
212
0.5 x 0.5
P7S1O
See Note*
1.2 x 0.6
36
212
0.5 x 0.5
P7S14
See Note*
1.5 x 0.6
39
212
0.5 x 0.5
P7S27
Sternum heating
Faint 1.5 x 0.6
-20
302
1.0. x 1.0
P7S35
Not visible
30.6
302
1.0 x 1.0
P7S50
1.5 x 0.8 x 0.8
1.2 x 0.6
34.5
212
0.5 x 0.5
P8S6
See Note"*
2.0 x 0.7
38.8
318
0.5 x 0.5 at z=12 cm
P8S14
See Note"
2.5 x 0.7
42.9
212
0.5 x 0.5 at z=10 cm
P8S18
See Note**
4.0 x 0.7
33.6
212
0.5 x 0.5 at z=9 cm
P8S22
See Note"
4.0 x 0.7
29.0
212
0.5 x 0.5 at z=9.5
P8S26
2.5 x 1.0 x 1.0
1.9 x 0.9
27.0
345
1.0 x 1.0
P8S32
These four sonications produced lesions measured post mortem but the corresponding sonication is unknown. The autopsy lesions
had cross sections of 0.8 x 0.8, 1.5 x 0.5, 1.5 x 0.8, 1.5 x 0.7 cm 2
the cumulative lesion dimension after
These four sonications were used to produced a single long lesion. The MR dimensions are of
3
each sonications. The end lesion size measured on autopsy was 4.0 x 0.8 x 0.6 cm
P4S71
On histological evaluation, the individual liver lesions could be divided into two
regions (see Fig. 6-36 and Fig. 6-37).
The outer region contains red blood cells and
decreasing density of hepatic cells as compared to normal tissue.
The inner region
contains small, dense nuclei (pyknotic indications of coagulative necrosis) and very few
red blood cells. There are also gaps between cells indicating a breakdown of intercellular
matrix. The borders between healthy tissue and lesion were visible to the eye but not as
precise at a cellular level. The histological results are consistent with those found for a
single focus transducer (Sibille et al. 1993; Vaughan et al. 1994).
150
Fig. 6-36: H&E stained tissue of the large thermal lesion of Fig. 6-34.
Lesion
Border
Normal
Fig. 6-37: Liver tissue stained with H&E. The left picture is of the tissue lesion with no vasculature
and pyknotic nuclei. The center picture is of the hemorrhagic border between viable and lesioned
tissue. The right picture is of healthy liver tissue.
151
Respiratory motion artifacts influenced the ability to extract clear temperature
images for some of the sonications. For example, Fig. 6-38 shows a time series of axial
temperature images taken during a large focus sonication. In some of the images the
heated tissue is not visible. This is most like due to the method of temperature imaging.
To measure temperature using the proton resonant frequency, a baseline image is taken
before the onset of sonication. Changes in the phase for each pixel are compared to this
baseline image.
The respiratory motion can cause inaccurate image registration and
degrade the temperature image. Occasional images during the sonication, however, do
align with the baseline image as the liver moves with the diaphragm leading to valid
temperature images.
The poor image signal-to-noise and motion artifacts made the
nondestructive temperature localization of a single focus unachievable
in the
experimental protocol.
Fig. 6-38: Series of temperature images which were highly affected by respiratory motion. The
images were taken in 6.6 s intervals from the onset of sonication (order is from left to right and top to
bottom). The image in the upper right corner corresponds to the end sonication temperature map
and an arrow has been place pointing to the focus.
152
6.3.5 Heating Comparison of Different Tissues
The 0.5 x 0.5 cm2 pattern was used for thigh, kidney, ex vivo / in situ liver, and in
vivo liver. The heating and cooling profiles differed significantly both in magnitude and
in duration. Fig. 6-39 contains sample profiles from the different tissue types. Notice
that only the kidney tissue temperature returned to its baseline in the two minute time
period after sonication.
50
W-
50
Thigh: 106 W
40
40
30
30
20
I-
In Vivo Liver: 212 W
20
10
10
0
1~
20
40
60
80
100
20
40
Time (sec)
60
50
W-
25
Ex Vivo Liver: 212 W
60 80 100 120 140
Time (sec)
Kidney: 212 W
20
40
15
30
10
20
5
10
0
(
20
40
60
Ao
100 120 140
0
20
40
60
80
100 120 140
Time (sec)
Time (sec)
2
Fig. 6-39: Temporal heating profiles of the 0.5 x 0.5 cm focal pattern in different tissues. Notice that
the peak temperatures differ significantly for the given power application.
153
6.4 Discussion
6.4.1 In Vivo Thigh Experiments
A total of 29 sonication lesions were produced in the thigh muscles including
twenty which were produced from the large focal volume sonication.
The thigh
experiments indicate that large (>5 cm 3), deep seated lesion can be created in vivo in a
single sonication, verifying the simulation studies by (Wan et al. 1996; Fan and Hynynen
1996a). While the lesion can be created in a 20 second sonication time, it would be
misleading to say that the tumor treatment rate is greater than 5 +20=0.25 cm 3/s since
there is a significant cooling time necessary before an adjacent sonication can occur
(more than 5 minutes). As demonstrated by simulations for phased arrays (Fan and
Hynynen 1996b) and similar to experimental results for single focus transducers
(Damianou and Hynynen 1993; Malcolm and ter Haar 1996; McDannold et al. 1998b),
the cooling time is necessary to avoid excessive near field heating and undesirable boiling
of water in the prefocal tissue. The MR temperature images demonstrate the long period
needed to cool near field tissue following a large focal volume sonication. The
overlapping sonications in the thigh, however, showed that more than 25 cm 3 could be
treated in less than 90 minutes. The interval between sonications used in this experiment
may have been longer than required.
No lesion was formed in the focal plane for three of the large volume, high power
sonications. This was predicted by the MR temperature images. In each case, a muscle
interface appeared to block the ultrasound transmission such that there was little or no
measurable temperature rise beyond the interface. The temperature images of Fig. 6-18
154
help explain the variable results. In the first sonication, the interface appears to have
caused some beam distortion (possibly refraction) and some heating near the interface
leading to a lesion with a "bent" shape with a wider region near the interface.
The
temperature image of the second sonication indicates a wide, diffuse focal region after the
ultrasound passed through the interface. The larger focal region led to a lower average
temperature and a less distinguishable coagulation volume. The third row of the images
indicate that the transducer side of the interface was heated to high temperature leading to
a pre-interface lesion. It is possible that cavitation was produced in the interface causing
the excessive temperatures and a significant decrease in energy transmitted beyond the
interface (a method to detect cavitation was not employed in this sonication). Interface
related cavitation events have been demonstrated before (Hynynen 1991; Hynynen et al.
1993).
In each case, the temperature images taken during sonication still accurately
predict the T2-weighted image measurements of lesion volume in the muscle although the
accuracy of the temperature images at the interface is compromised by the lack of
temperature dependent proton resonance shift in fatty tissue.
The experimentally measured temperature and thermal dose response measured
using the MR scanner correlated very well with the simulated temperature distribution for
the large focal volume sonications. The same simulated and experimental power level
yielded almost identical peak temperatures. The difference between the simulated and
measured temperature during cooling is most likely a result of the simulated perfusion
being too high (1 kg/m 3/s). In the case of single focus sonications (such as the axial
scanning presented in this paper), there is a significant difference (up to a factor of three)
between the power simulated and experimentally tested to produced the same temperature
155
rise. This is similar to the results found in dog thigh muscle (Moros and Hynynen 1992)
which indicated that the peak intensity of a single focus beam was more highly attenuated
than the total power of the beam. This was attributed to possible scattering and refraction
of the beam. The large focal volume sonications in this study did not demonstrate this
same result and therefore the model used in this study performs better for large focal
volumes than the small volumes produced by single focus transducers.
This is in
agreement with the results from Fjield (Fjield et al. 1998).
The in vivo thigh muscle experiments also demonstrate that a well constructed and
driven phased array transducer can be accurately controlled without invasive acoustic
feedback measures at clinically significant depths. Lesions much larger than the volume
of a single focus lesion were produced at depths of 8 cm below the skin. This is deep
enough for almost all extremity treatments, breast tumor treatments, and kidney
treatments, and it would be enough for some liver mass treatments.
6.4.2 In Vivo Kidney Experiments
The in vivo kidney sonications prove that the 256 element array can be used to
create large lesions in tissue that has high blood perfusion. This is important since some
tumors could exhibit perfusion higher than the non-pathological tissue (Bowman 1984;
Bowman 1986; Bowman et al. 1995). Tissues with high perfusion have generally been
more difficult to heat due to the heat carried away from the focal region to the rest of the
organism. By using a relatively short sonication (20 seconds) the effects of perfusion are
minimized in the heating stage of the treatment but are still significant for large focal
volumes due to the extensive cooling period (which may last from one to ten minutes).
156
While knowledge of the perfusion rate is critical for some treatments, the MR
temperature images help alleviate the need for specific measurements of perfusion
because temperature images and corresponding thermal dose estimates can be made for
target tissue volumes.
6.4.3
Ex Vivo/ In Situ Experiments
The sonications into the cadaver of a pig illustrate that MR can be used to
evaluate acoustic apertures and detect undesirable heating in ribs and other obstacles.
Unlike traditional X-ray or CT scans, the SPGR MR sequence can clearly distinguish not
only dense boney obstacles, but also acoustically attenuating low density cartilage. It is
important to note that the liver is predominantly covered by some kind of acoustic
obstacle: costal (bone) ribs on the posterior and lateral side of the body, chondral
(cartilage) ribs on the anterior side, lung tissue on the superior/posterior edge, and
occasionally bowel and/or stomach from the inferior/anterior side. The exact location of
these obstacles varies from patient to patient as it did from pig to pig in these
experiments. A method to evaluate the location of available acoustic windows is critical
for treatment operation and is well suited for the 3D image acquisition available through
MR imaging.
Some of the acoustic obstacles can be eliminated through simple procedures. For
example, the stomach can be deflated through a nasogastric tube. The parts of liver
blocked by lung may be cleared by using an end expiratory sonication although the
superior motion of the diaphragm would tend to move the liver further underneath the
ribs.
It has been proposed that the liver or other tissues could be treated by using a
157
phased array to sonicate between the ribs or bones, but this has not been experimentally
tested (McGough et al. 1996; Botros et al. 1997; Botros et al. 1998).
The MR
measurement of the temperature rise in the intercostal rib space would be valuable to
evaluate these treatment modalities but future work to quantify this temperature rise in
bone and cartilage would be necessary. The ability to measure temperature at the rib
level could make it possible to sonicate the liver through the low density cartilaginous
ribs. This could significantly increase the available acoustic window for liver treatment.
Lastly, the results from this research indicate that any therapy that does induce rib heating
will suffer from extensive cooling periods to avoid near field heating between ribs.
6.4.4 In Vivo Liver Experiments
To the author's knowledge, this research is the first application of a continuous
wave ultrasonic phased array used to produce thermal coagulation in vivo liver and kidney
of a large animal model. The in vivo tests indicate that a single 20 second sonication can
produce lesions greater than 1-2 cm 3 in the liver through the use of phased array
applicators. This research also showed that long overlapping lesions from multiple focus
fields can be created in the liver. The in vivo kidney sonications prove that the ultrasound
array can be used to create large lesions (>0.5 cm 3 ) in tissue that has high blood perfusion
using a single sonication. This is important since some tumors could exhibit perfusion
higher than the non-pathological tissue. Tissues with high perfusion have generally been
more difficult to heat due to the heat carried away from the focal region to the rest of the
organism. By using a relatively short sonication (20 seconds) the effects of perfusion are
minimized in the heating stage of the treatment but are still significant for large focal
158
volumes due to the extensive cooling period (which may last from one to ten minutes).
The MR temperature images help alleviate the need for specific measurements of
perfusion because temperature images and corresponding thermal dose estimates can be
made for target tissue volumes.
This research also showed that liver tissue can be geometrically targeted using
presonication MR SPGR or T2-weighted images, that the heated liver tissue can be
monitored during sonication and the cooling period by measuring the proton resonant
frequency shift, and that post sonication T2-weighted images can be used to detect liver
lesion volumes for lesions formed with high temperatures (but below boiling or cavitation
thresholds).
One aspect of the experimental MR protocol that could further improve treatment
monitoring of the liver in a patient is the reduction of respiratory motion artifact.
Although accurate targeting and sufficient monitoring could be achieved in a breathing
pig for most sonications, the motion artifact degraded the MR images significantly. The
image degradation is well known in the MR field, and patient studies have quantified the
motion. It has been shown that the liver has predominantly a superior/inferior motion
during quiet respiration with an average displacement of 1 cm although deep respiration
can be 10 times that distance (Davies et al. 1994).
The image degradation could be
avoided by using a breathe hold protocol for the patient or by regulating the breathing
through a ventilator. The prior option is advantageous because it would still not require
general anesthesia; the latter option is advantageous because the location of the liver
could be more accurately positioned. The ex vivo tests indicated that motion free images
could be used to non-invasively detect low temperature subnecrosis sonications in the
159
liver. From a clinical standpoint, control of the patient breathing could help eliminate the
ill effects of sonicating a moving tissue volume (Vaughan et al. 1994; ter Haar et al.
1998a). While theoretically the array developed in this research could electronically track
the 1 cm displacement produced in the liver during quiet respiration, the coordination
between electronic focusing and liver localization is complicated. A better method would
use a large focal volume in a breath hold technique to ensure that the heating at the focal
volume was sufficient to coagulate tissue despite the fact that the liver did not return to
the exact position after each respiration. Overlapping sonications could then be used to
cover the entire target tissue volume with less chance of leaving small volumes of
surviving tissue which could be missed by overlapping a small single focus.
The second method to improve MR images would be the use of a more complex
MR signal receiver. The RF gain from a circular MR surface coil drops off significantly
at distances on the order of the coil aperture. This indicates that a large bore coil should
be used for deep sonications. Unfortunately, the overall gain of a coil also decreases with
aperture size. Therefore, for deep sonication in a patient, it is recommended that a more
sensitive signal acquisition receiver such as a MR array coil be developed.
6.5 Conclusion
This research has shown the feasibility and advantage of using a large scale
ultrasound phased array in vivo for MR guided ultrasound coagulation of large volumes
of deep seated tissue. The phased array offers a control and flexibility not available in
single focus transducers or in less numerous arrays. The thigh experiments have shown
that the theoretical model used in this research can accurately model the response of large
160
focal volumes when large interfaces are not close to the focal volume. This study has
also confirmed the ability of MR to detect lesion forming temperature elevations and
determine treatment effectiveness post sonication for lesions at clinically significant
depths. In a clinical setting, the advances in control will make patient treatment more
accurate as well as more clinically feasible.
This research has also shown the feasibility of using a large scale ultrasound
phased array for ultrasound surgery in the liver of an in vivo large animal model. It has
demonstrated the importance of temperature monitoring using MR to detect high
temperature elevations and post-sonication lesions in the liver. MR can be used to avoid
and detect heating in acoustic obstacles such as cartilage and bone. Using MR guided
surgery with a phased array applicator could greatly improve the quality of a treatment as
well as decrease its treatment time.
161
7. CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK
7.1 Conclusions
The use of high power ultrasound to non-invasively treat deep seated tumors is
rapidly moving towards becoming a clinical procedure.
Currently there are phase 1
clinical trials of single focus transducers in Japan, Britain, China, Canada, and the United
States for the treatment of breast, liver, kidney, and prostate disorders. These trials all use
fixed focus transducers with ultrasound guidance and no thermal monitoring (the
exception being the MR guided breast treatments in Canada and in the Focused
Ultrasound Surgery laboratory associated with this thesis).
While the few published
preliminary results are promising using only image guidance, the need for treatment
monitoring and larger focal volumes is well documented. This study has theoretically and
experimentally shown that a large scale phased array can be used in conjunction with MR
guidance to coagulate clinically significant tissue volumes in a single sonication.
In
particular, the system was used in vivo in a large animal liver model to demonstrate the
feasibility for a future human trial.
This study developed a sophisticated hardware design for driving therapeutic,
continuous wave arrays. The system implemented automatic phase and power feedback
to ensure proper driving signals to the array's elements. It demonstrated (both in acoustic
tests and in vivo experiments) that the acoustic output of a well built array can be
significantly improved without array specific calibration or invasive hydrophone probes.
162
This study determined techniques and materials adequate for constructing a large
scale phase array. These included extensive material tests of both diced PZT arrays and
1-3 piezocomposite materials.
It was found that 1-3 piezocomposite materials can
generate output powers high enough for the clinical treatment of deep seated tissue
despite significant interelement mechanical coupling. By using the composite materials,
the drawbacks of diced arrays such as the lack of a continuous ground plane, weakened
structural support, limited element size, and complex water proofing can be lessened or
eliminated. The use of an acoustic matching layer with a 1-3 piezocomposite material
was found unnecessary for the two arrays tested in this study. Overall, two 16-element
arrays, a 64-element array, and a 76-element array were created with the "dice and fill"
method. Eight 4-element, a 9-element, and a 256-element array were constructed from
1-3 piezocomposite materials.
A theoretical study was presented to determine the maximum size of necrosis for
a phased array transducer. The theoretical model demonstrated that the limiting factor in
large volume coagulation was the extension of heating into the near field between the
array and the target tissue volume. A protocol based on limiting the thermal elevation in
the prefocal tissue was determined according to the pain thresholds for temperature stated
in literature. The theoretical maximum focal volume using this technique was found to
have about a 2 cm diameter for most array designs. This study can be used a starting
point for basic therapeutic array design.
Experimental tests to coagulate tissue were performed using temporally switched
acoustic fields from a constructed 16 element array in an in vivo rabbit thigh muscle.
These tests were the first experimental demonstration of thermal coagulation using
163
temporally switched fields. MR imaging was used to map thermal fields, and techniques
to optimize thermal dose using a large multiprocessor computer were developed. The
field optimization technique was shown to decrease the average treatment power and the
theoretical treatment time for a large tumor volume. The theoretical studies indicated that
treatment variances due to tissue inhomogeneities can be decreased if the switching cycle
of the acoustic patterns is sufficiently fast.
Lastly, a 256 element phased array was used in a series of in vivo porcine
experiments to demonstrate the control and advantage of a large scale phased array for the
treatment of deep seated tissue. The array could coagulate large volumes of tissue in
thigh (>5 cm 3), kidney (>0.5 cm 3), and liver (>2 cm 3) using a single 20 second sonication.
7.2 Recommendations for Future Work
The main target disease for this research has been the treatment of liver tumors.
Other organs such as prostate (through the bladder), breast, pancreas, and brain may also
benefit from the use of extracorporeal large scale arrays for the coagulation of tissue.
However, the recommendations for future work will focus predominantly on liver
treatment.
The system developed in this thesis is an adequate prototype for animal
experiments, but changes to several aspects of the current set up could make the MR
monitored treatment of human liver tumors more effective.
First, the array geometry
could be designed to accommodate the available acoustic window to the human liver.
This could include the construction of oblong rather than circular aperture arrays for
treatment between ribs or a triangular aperture array to match the acoustic window
164
inferior to the sternum.
Second, while the pseudoinverse technique to determine the
driving signals for an array was found adequate for small numbers of foci, a more
effective technique to optimize the thermal dose field using a large number of
simultaneous foci could further improve treatment operation.
The temporal switching
technique developed in this thesis helped alleviate the problem, but new, less
computationally intensive methods to optimize the field distribution would be valuable
for real time treatment.
Third, techniques to reduce the physical array dimensions
(mainly depth) would be helpful in the space limited bore of an MR scanner.
From a clinical perspective, this study recommends that the treatment of liver
tumors be performed with a patient breathe hold. This could be done under voluntary
patient control or through the use of a mechanical respirator. There are several reasons
that this would improve the treatment: 1) the target tissue could be more accurately
delineated in the presonication MR image, 2) the thermal imaging would not suffer as
much from motion artifacts that can distort the ability to monitor the treatment, 3) the
coagulated focal volume would be more easily formed since the thermal elevations could
be localized at a stationary position rather than a moving target, and 4) multiple
overlapping focal volumes could be more accurately placed for the treatment of complete
tumors. Although the 20 second sonication demonstrated in this thesis is short enough
for a patient controlled breath hold application, it is the opinion of the author that
respiratory control using a ventilator would yield more repeatable results and should be
used if the patient can undergo general anesthesia.
Lastly, several MR techniques need to be implemented before the full potential of
a MR guided ultrasound phased array surgery can be realized. First, surgical treatment
165
software including 3-D segmentation techniques to automatically localize the liver in
relation to both boney and cartilaginous ribs is needed to avoid acoustic obstacles for real
time surgical procedures. Second, new imaging sequences to eliminate motion artifacts
from the temperature images could greatly improve the ability to monitor liver treatments
(even if respiration is controlled, there is still a small motion artifact from the beating
heart). Third, it is recommended that the ultrasound array be designed in conjunction
with MR coil technology such that the MR coil is optimized to monitor the signal
intensity at the focal tissue depth of the ultrasound device.
development of MR coil arrays for deep seated sonications.
166
This could include the
8. APPENDIX A: ULTRASOUND DRIVING SYSTEM DOCUMENTATION
8.1 Introduction
This appendix is offered as a more detailed description of the phased array
ultrasound driving hardware system described in Chapter 2. This appendix will cover the
basic system architecture and circuitry implemented to drive phased arrays of multiple
sizes and shapes. This documentation will focus on the analog portion of the individual
driving system channels. The embedded software and communications techniques will
be briefly discussed but, as they were not prepared by the author of this thesis, they will
be documented in other reports.
This appendix assumes little RF design knowledge from the reader since it was
prepared for an audience of therapeutic ultrasound researchers. Parts of the discussion,
therefore, may be simplistic for a reader associated with more complex RF design
principles.
The documentation will attempt to explain in a simplified manner the
operation of the system while the references can offer more detail and circuit theory.
8.2 System Block Diagram
Fig. 8-1 is a picture/block diagram of a completed ultrasound driving system. The
system is based on distributed control. The user interfaces with the system using a PC.
Software such as a hyperterminal window or custom written applications (Visual Basic or
Visual C) along with associated .dll files have been written for the PC (a laptop in the
photograph) to communicate with a embedded single board computer (x486) found in the
167
control box of the system. The embedded computer directly communicates with the
ultrasound driving cards as well as controls the system frequency generator,
enables/disables the 48 V DC supplies, and interfaces with emergency shutdown
switches.
PC Control
Control Box
x486 Embedded Controller
Frequency Generator
48 V DC Supplies
Ir
Ultrasound
Driving Cards
Matching Cards
Array
Fig. 8-1: Photograph and block diagram of ultrasound driving system.
The ultrasound driving cards communicate with the embedded controller through
a serial bus from the embedded computer to the ultrasound driving card racks.
The
address of each card is determined by the rack address (set with a rotary switch on the
rack backplane) and by the hardwired location of the card in that rack. The system can be
expanded to 16 racks of 16 cards for a total of 1024 driving channels (4 channels per
card).
The RF output of the ultrasound driving cards is fed to a second rack of printed
circuits cards, each with a simple inductor/capacitor circuit to electronically match the
168
array elements to 50 I.
Each array must be matched with a set of these cards for the
system to operate properly. The array is then attached to the output of the matching cards
through a set of 64 line coaxial cables (cables on the right of the photograph).
The fundamental analog unit of the system is the ultrasound driving card and this
will be the focus of the rest of this appendix.
8.3 Ultrasound Driving Cards
Fig. 8-2 shows a photograph of the ultrasound driving card. Each card contains
the digital and analog circuitry to drive four 50 fl loads (or ultrasonic elements which are
matched to 50 fl). The right side of the card (furthest from the edge connector) contains
most of the digital circuitry: the 68HC 11 microcontroller, memory, addressing chips,
DACs, ADCs, and the phase locked loop circuitry. The left side of the card contains most
of the analog circuitry: voltage switching regulators (connected to the bar heat sink) and
their corresponding passive circuitry, switching FETs on back of card, filter components
(large toroids), and dual directional couplers (small upright toroids). The analog circuitry
for the first channel on this card is found in the upper right corner of the analog sections.
The other channels proceed in numerical order counter clockwise. Appendix B contains
the complete schematic of the card.
169
Fig. 8-2: Photograph of an ultrasound driving system card (front view on top, and back view on
bottom).
170
Fig. 8-3 contains a block diagram of the ultrasound driving card. This differs
from the block diagram found in Chapter 2 since it only contains the subunits which
appear on the printed circuit board. The inputs to the card include the power supplies
(+5,-5, +7, +48 V), digital frequency clock (running at 16 times the RF signal output), RF
voltage from the transducer side of the matching cards, and the communications signals
to the MCU from the embedded controller. The output signals include the high power RF
sinusoid and communications measurements to the embedded controller. Each of the
subunits in Fig. 8-3 will now be detailed.
Communication Commands
Addressing
MCU
Memory
ADC
48 V supply
Peak
Detector
DC-to-DC Down
Converter
Common
Clock
Phase
Shifting
Phase
Correction
DC-to-RF
Conversion
RF
Filtering
Power
Coupler
-
RF Out
Phase
Detector
Phase
Detector
Transducer
Voltage
Fig. 8-3: Block diagram for the ultrasound driving system card.
8.4 System Subunits
The subunits can be divided into three categories: power, phase, and control. The
power category is responsible for generating and regulating the high power sinusoidal
voltage output. The phase category is responsible for generating and regulating the phase
171
of the output sinusoid. The control category is responsible for establishing the set points
of voltage and phase while monitoring the circuit to verify proper operation.
This
appendix will describe the analog circuitry of the power regulation loop and the phase
correction loop.
8.4.1 Power Loop
8.4.1.1 Overview
The power category contains the high power DC-to-RF converter, the supply DCto-DC down converter, the output RF filtering, and the power measurement components.
The circuit architecture was chosen to produce a high DC-to-RF efficiency such that the
amount of heat sink bulk was minimal. However, the circuit was not optimized to reach
its ideal efficiency although a significant amount of bench work was spent adjusting the
Fig. 8-4 contains a copy of the power loop
circuit operation to appropriate levels.
circuitry from the schematics in Appendix B (Channel #4 on card).
C50
C49
0.01pFTl
3.31&F
TI11
Q10
107
+4
0.19 F
C.5
AC 401F
1R2
E
C
DL4148
D25
C52
IRF634S
200pF
L13
L14
10.2pH
7.4pH
T2D
10:10
7.
"2
.1pF
FT50-77
2400pF48
C53
54
112
9D
00
1
C136
RFout4
1F
m450
74
LT107
V
o50
D28 DL4148
R4LD27
D-4148
C59
C57
D30
674
C58
D29 DM2U148
L38
22pH
2K
>U53A
Vrefl4
R 39
C2K
Fig e
-4:
C03p
1wer
r41
0
C61
.IP01
LMC6484
D32 D148
R4p
DIA18
C6
100-If2K
UB
1143
look:
C64
0.01lpF
7Vfor4
LMC6484
CH4Poer SetLM64
Fig. 8-4: Complete power ioop circuitry (from Channel 4 of schematics in Appendix B).
172
8.4.1.2 DC-to-RF Conversion
8.4.1.2.1 Topology
A schematic of the DC-to-RF converter is found in Fig. 8-5. The purpose of this
subunit is to convert a digital 0-5V RF square wave at the ultrasound driving frequency
into a high power sinusoid from a DC supply. The converter was designed to be a
combination of the class D and class E amplifier in a push-pull topology. The two active
switches (IRF634S FETs) are switched out of phase by the FET driver IC MC34152. The
high pass filter circuitry and diode between the FET driver and the FET prevents
Drain to source
continuous FET conduction in the case of a lost RF clock input.
capacitors are placed in the circuit to create signals consistent with a class E amplifier.
The DC voltage supply is coupled to the switches through a large inductor and a center
tapped transformer to act as a DC current source for the active switches. The AC drain
voltages are transmitted to the RF output through the center tapped transformer.
From DC Supply
+7V
C50
C49
3.3pF
0.01pF
Ull
RF CLOCK4
Q10
2 C51
OAPF
.
U2AA
L4R34
p DL4648
200pF
T1O
10:10
MC34152
200pF
S634S
74AC04
FFle
Filter
To IlF
C2T
C52
I341342
0.
K
DL4148
Shutdown4
1005D
48
Fig. 8-5: Schematic of the DC-to-HF power converter.
173
Class D and class E amplifier are discussed extensively in (Baxandall 1959; Senak
1965; Raab 1973; Sokal and Sokal 1975; Raab 1977; Murray and Oleszek 1979; Sokal
1981; Kazimierczuk 1984; Granberg 1985; Kazimierczuk 1986; Avratoglou et al. 1989;
Kazimierczuk and Tabisz 1989; Kazimierczuk 1993; Sowlati et al. 1995; Koisurni et al.
1996; Sowlati et al. 1996). The basic principle of these amplifiers is the use of an active
switch which is driven on and off at the frequency of the signal to be created. A class D
amplifier typically is characterized by a two pole switch (two FETs) with bandpass or low
pass filtering to produce a voltage or current sinusoid. A class E amplifier can be defined
as a subset of switching amplifiers that fulfill the following: 1) the voltage rise across a
transistor drain is delayed until the transistor is off, 2) the voltage across the transistor
drain is zero when the transistor is turned on, and 3) the slope of the drain voltage of the
transistor is zero when the transistor is turned on. Raab calls the circuitry which meets
these requirements exactly an "optimal class E" and he calls a circuit which falls short of
these requirements a "suboptimal class E" (Raab 1977). The circuitry presented in this
appendix can be classified as "suboptimal class E."
The basic class E amplifier introduced by the Sokal's is found in Fig. 8-6 (Sokal
and Sokal 1975).
The circuits consists of a large inductor (LI) which converts the
voltage source (Vcc) into a DC current. The transistor (TI) is used as a switch with its
drain capacitance paralleled with an external capacitor (Ci) to eliminate fast voltage
spikes. The tuned circuit (C2-L2-R) is determined to guarantee that the resonance of the
overall circuit is damped such that the class E requirements are met.
The switching
frequency is chosen to be slightly higher than the resonance of the tuned circuit assuming
a high
Q.
The component values can be determined either numerically or through some
174
simple equations if assumptions are made about the tuned circuit bandwidth. Equations
for the choice of element values are found in references for Class E amplifiers.
Vcc
Li
C2
L2
Ti
Fig. 8-6: Basic class E amplifier.
The circuit used in the ultrasound driving system is a variation of the basic class E
amplifier.
The most obvious modification is the implementation of a push pull
architecture. Raab suggests the implementation of a push-pull class E amplifier using an
individual DC choke current source for each switch and a transformer to couple the
switches to the tuned circuit load (Raab 1977). The ultrasound driving system in this
research utilizes a single current source inductor to feed both FETs through a center
tapped transformer.
8.4.1.2.2 Components
The IRF634S Power MOSFET was chosen for its large drain-to-source
breakdown voltage (250V), high drain current rating (5.1 A at 1000 C and 10 V), its low
drain-to-source on resistance (0.45 fl), medium output capacitance (190 pF), and its fairly
fast turn on and turn off time (20 ns).
It also has an integral reverse diode with a
maximum forward voltage of 2.0 V to clamp the signal of the push-pull topology.
The shunting drain capacitors were chosen to be surface mount mica capacitors
for their low loss (Q>6000) and small lead inductance.
175
The transformer was constructed from a multiaperture core of a Nickel zinc
material (material 43 Fair-Rite) with 24 AWG magnet wire (10 turns on secondary and 5
turns on each primary winding). It was chosen for its relatively high permeability up to 2
MHz (850) and relatively low parallel resistance at 1.5 MHz. The core size was made as
small as possible although it led to difficulties in transformer construction.
The DC current supply inductor was a high current, surface mount inductor from
Vishay Dale (maximum DC resistance 0.12 ft rated at 2.8 A).
Motorola's MC34152 high speed dual MOSFET driver IC was chosen for its fast
rise and fall times for capacitive loads (15 ns) and an AC logic inverter was used to
produce sharp edges for the driver.
8.4.1.2.3 Electrical Waveforms
Fig. 8-7 shows a typical drain voltage and the corresponding secondary current for
a 1 W RF output to a 50 fl load at 1.5 MHz. Notice that the circuit operation is not a
pure class E response since the voltage and its slope are not at ground levels in the switch
turn on time. This overdamped resonant circuit presents as a "mitten" shaped waveform
(the "thumb" the mitten is the upward voltage ring). This is similar to the response at 60
W (Fig. 8-8) although the relative power loss is less. Similar plots at 1.1 and 1.8 MHz
are found in Fig. 8-9 and Fig. 8-10. From these plot, one can see that proper class E
operation occurs at a frequency between 1.5 MHz and 1.8 MHz (at 1.8 MHz, the tuned
circuit is underdamped). A plot of the FET gate and drain voltages at 30 W are found in
Fig. 8-11 through Fig. 8-13. The transformer primary center tap and secondary voltages
are found in Fig. 8-14.
176
SOOMS/s
Tek E
623 Acqs
T I.....
.......
....
... ... ... ... ... .. ... ......
....
. ... ... ... ... ..... ........
....
... ... ..
....
. ... ... ... ... .... ..... ....
... .. ... ......
....
. ... ..
.. ... ... .... ..... ....
....
.. .... ... ... ... .. ... .....
wE
....
. . . . . . . . . . . . . . . . . . . . . . ....
....
. . . . . . . . . . . . . . . . . . . . . . ...
24 .....
. . . . . . . . . . .
. . . . . . . . . ....
....
. . . . . . . . . . . . . . . . . . . .
....
. . . . . . . .
. . . . . . . . . . . . . . . . ....
&Wjj
5V
Ch2
10UmV
M 10uns
. . . . . . . ...
Cli
-L.
13 V
14 Aug 1998
17:18:51
Fig. 8-7: FET drain voltage and transformer secondary current for a 1 W RF output signal
(2mA/mV for current on Ch 2).
Tek
SOOMS/S
331 Acqs
..............................
......................................
.................
..........
...............
....
..........................
....
..........................
....
............... ..........
..... ....................
....
...
......................
............
.........
....
........
....
.......
.......................
I....
....
!..L'
....
..........................
....
..........................
am1
soy
Ch2
50 V
..............
M 1O0ns Ext %
1.5 V
14 Aug 1998
15:58:25
Fig. 8-8: Superimposed FET drain voltages at 30 W RF output (1.5 MHz). In this case, the push-pull
architecture is not very well matched as is evident from the different peak voltages across the FETs.
177
50OMS/S
Tek
1841 Acqs
.............. ...............
.............
.................................
-T-
...
. . .. . . . . .
. . . . . . . . . . . . . ....
...
. . . . . . . . . . . . . . . . . . . . . . . ....
...
..........
...
........
..
. . . . . . . . . . . . ....
...
. . . . . . . . . . . . . . . . . . . . . . . ....
. . . . . . . . . . . ....
...
. . . . . . . . . . . . . . . . . . . . . . . ....
EJ
....
..........................
....
..........................
....
..........................
....
.........................
..........
...
Lqw
SU V
U12
50 V
m i0ons
EXt
-L
1.3 V
14 Aug 1998
16:00:21
Fig. 8-9: FET drain voltages at 30 W RF output at 1.1 MHz.
TekOM SOOMS/s
1259 Acqs
..........................
.......................................
T............
I
. . . . . .. .. .
.. . . .. .. . .
.. . . .. .. . .
El
.................... .
........................
Am
5U V
Ch2
50 V
M IOOnS
EXt
X
1.5 V
14 Aug 1998
16:02:19
Fig. 8-10: FET drain voltages at 30 W RF output at 1.8 MHz.
178
409 Acqs
Tek@M SOOMS/s
.................
..............................
......................................
.................
I ...........
...
........................... .................
....
......... ................
...
. .........................
..................... ...................
........................... .........................
...
................
...
....................
Lim
5 V
Chi
So V
M i0ons
Ext
%
1.5 V
14 Aug 1998
15:50:01
Fig. 8-11: FET gate and drain voltages at 30 W RY output at 1.5 MHz.
368 Acqs_,
2GS/s
Tek
11
F.................................... :
a...........
................................................
: ......................
j
LM
....
.
.
.
.
.
.
....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. ...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
....
.
.
.
.
.
.
.
.
.
.
....
..
.
.....
......
...................... ..
...
..........................
........................
...
...................
......
..........................
2-+ ....
5 V
Ch2
3 V
M
25ns
Ext
X
196MV
14 Aug 1998
13:19:30
Fig. 8-12: FET drain and gate at FET turn on time (1W, 1.5 MHz).
179
Te
887 Acqs
kjst3 2GS/s
.T...............
w.
.4
5
. .........
i.
h2
5V
M
25ns
Ext
I
196mV
14 Aug 1998
13:18:13
Fig. 8-13: FET drain and gate at FET turn off time (1 W, 1.5 MHz).
TekhtWU SOOMS/s
..
..........
M
1291 Acqs
T ......I .................
................ ....
....
..
..
....
..
. .............. . ......
..............
.....
...
..
.....
..........
....
....
........................
...
..........................
. ..............
... . . . . . . . . . . . . .
2-+
... . . . . .
. . . . . . . . .
.... . . . . . . . . . . . . . . .
. . . . .
.... . . . . . . . . . . . . . . . . . . . . . . ...
. . . . . . . . . . .
... . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . .
. . . . . . . . . . . . . . . . ...
. . . . . . . . . . . . . . . . . .
Lem
100
v
C12
100
v
M IOons
Ext
I.
1.5 V
14 Aug 1998
16:08:47
Fig. 8-14: Voltages at the transformer primary center tap and secondary / filter junction for a 60 W,
1.5 MHz signal.
180
8.4.1.3 DC-to-DC Down Converter
8.4.1.3.1 Topology
A DC-to-DC switching converter is used for each channel to scale the system's
48 V DC supply down to a lower DC level as a source for the DC-to-RF converter
described previously. This converter is based on the LT1074 (Linear Technology), 100
kHz voltage switching regulator in a buck configuration. The converter operates in both a
discontinuous and continuous mode depending on the desired RF output power (the
breakpoint is at about 20 W). The voltage sense feedback signal originates from the RF
power measurement circuitry (described later) and is compared to the power set point
through an operational amplifier integrator. A schematic of the DC-to-DC converter is
found in Fig. 8-15.
LT1074
+48V
5
C59
C
0.01 pF
U12
_
4
I7
I
L15
47pH
57220p
C64
Fig.
C60i JC58ToLd
220pF
0.01 pF
D30
8-15:
MC63
10
Dtf
100
O4
PCw4
Poe
eebctSga
Fig. 8-15: DC-to-DC buck converter schematic for Channel 4 on UDSC.
181
For the novice, a number of excellent texts exist on switching converters. Two of
these are: Power Switching Converters by Ang and Switching Power Supply Design by
Pressman (Ang 1995; Pressman 1998). In addition, Linear Technology has published a
number of application notes are useful (Application Notes 25, 35, 44, 46).
A basic diagram of the buck converter operation is found in Fig. 8-16.
The
converter uses a switch which is opened and closed at a fixed frequency but with varying
duty cycle.
The inductor, capacitor, and load act as a filter such that a DC voltage
appears at the load.
The duty cycle of the switch is regulated to control the output
voltage, output current, or other parameter from the circuitry (in the ultrasound driving
system, the switch duty cycle is regulated by the measured RF forward power delivered to
the load). When the switch is closed, the DC voltage (VDC) appears across the reversed
clamp diode (Dl). Assuming the load is being regulated at a constant DC voltage, current
flow through the inductor (LI) begins to flow towards the load and capacitor (Cl). If LI
is chosen appropriately large for the switching frequency (and the capacitor/load shunt),
the current increase is linear with time. When the switch is then opened, the current
magnitude through the inductor begins to fall linearly as a function of the new voltage
across it (voltage is clamped to the diode drop on left and the load voltage on right). If
the inductor current does not fall to zero (at which point the clamping diode reverse
biases and clamps off the current), the buck regulator is running in continuous mode (for
continuous current). If the current through the inductor does fall such that the current is
cut off by the reverse bias of the clamp diode then the regulator is running in the
discontinuous mode.
182
LI
SW
Load
VDC
C1
Fig. 8-16: Symplified schematic of a buck converter.
Most switching regulators are designed to operate at a fixed output voltage with
variable current depending on the load impedance.
However, the regulation of the
ultrasound driving circuitry is designed to fix the output DC voltage such that the DC-toRF converter delivers the desired RF power. Therefore, this regulator was designed to
step down the 48 V input supply voltage from 0 V to 45 V depending on the desired
output power.
The advantage of using a switching regulator instead of a linear regulator is its
high efficiency. The losses in the DC-to-DC convertion can be attributed to losses in the
conduction current through the forward biased diode (sometimes called a DC loss),
resistive losses in the inductor and capacitor (sometimes called AC loss), and losses in the
active switch and DC supply. The disadvantage of the switching regulator compared to
linear regulators is increased noise commonly presenting as an AC ripple on the output
voltage level. The AC ripple on the DC output voltage is a combination of two factors:
the ac current being stored as a voltage on the capacitor (C1) and the ac current flow
through the equivalent series resistance (ESR) of the capacitor.
As the ripple voltage
could be contained to a small level, the swtiching regulator was chosen for this design to
produce a more efficient and compact power regulator.
183
8.4.1.3.2 Components
The LT 1074 voltage switching regulator has a 100 kHz switching frequency and a
maximum 60 V input voltage rating. It can be configured in a buck configuration with
output voltage between 2.5 and 50 V. It is rated at 5 A and therefore can supply the
necessary power for the ultrasound channel. The regulator is also fairly cheap and there is
plenty of data about it. Linear Technology's Application Note 44 is very helpful and one
of the examples in it was used as a starting point for the power regulation circuitry.
The flyback or clamp diode (D30 in Fig. 8-15) is a Schottky barrier rectifier. It
can chosen for its ability to deliver up to 3 A of continuous current and 100 A or surge
current. It also can handle up to 60 V of reverse bias (for the 45 V pulses of the VSR).
The filter inductor (47 jiH) is a pre-wound C&K pot core inductor that is rated for
3 A and a maximum DCR of 0.022 fl with minimum ET of 90 V s.
The filter capacitor (C58 of Fig. 8-15) is a radial lead 220 pLF electrolytic
capacitor rated at 50 V and 890 mA rms ripple current at 100 kHz. The estimated ESR is
0.3 fl.
C62 and R42 are a frequency compensation network for the VSR. These are set
to avoid large overshoot of the VSR at start up.
Their values were taken from the
suggested values in the Linear Technology application notes (see App Note 25 and 44 for
a discussion on how to empirically determine appropriate values). The op amp integrator
circuit is part of the feedback network for the power control. The feedback diode (D3 1)
is used to protect the VSR from high voltage inputs from the feedback op amp and to
compensate for the internal diode drop of the VSR (see Linear Technology Application
Note 44).
184
8.4.1.3.3 Waveforms
Fig. 8-17 contains typical voltage waveforms for the buck regulated DC voltage
and the output voltage of the switching regulator. The buck regulator changes from
discontinuous mode to continuous mode at about 20 W. Fig. 8-18 contains the AC output
ripple for various powers. The ripple is less than 2% for all powers. The filter inductor
AC current is plotted in Fig. 8-19. At 20 W RF power, the AC ripple current is 2.5 A (pp) which appears as a 200 mV (p-p) output voltage ripple indicating an ESR of about
0.1 fl. The DC current was not directly measured but for the 1 W and 20 W RF power
levels it is estimated to be 0.26 A and 1.25 A respectively (from the current graph in
discontinuous current mode).
This indicates that the DC-to-RF converter presents as
about a 20 fl DC load to the DC-to-DC converter.
TkEEE sOMs/s
1199 Acqs
(a)
TekEE
4
-
2+--
+ -.4+ +
[+-
{--T-*-
+
7'
--
T+
-
+
nz 2V
TekEEE2OMS/s
2oMS/s
540 Acqs
.(b
.....
2+
--
1
-+
M Is
1 4 Aug 1998
14:42:4
1544 Acqs
--
- +
-
---
- ---
oV
TekUEE20MS/S
-
--
M
-
+ -
%2"
'
6
174 Acqs
1 4 Aug99
15:36:9
(d)+
(c)
...t
............4 .....
....
....U
t
2-.
MUI
z2V
M1
2'
M . 5jas
Ih2
20.8
V
2
20
14 AUg19
15:34.:449
Ch2
20VM
25PS Ch2
20.9 YV
19
14 Aug 1999
15: 33: 11
Fig. 8-17: Regulated DC voltage (CH 1) and VSR output voltage (CH2) required for RF I owers of
(a) 1 W, (b) 20 W, (c) 30 W, and (d ) 60 W. Note that the vertical scale for CH1 chan ges. The
appropriate voltages were 5.5 V, 24 V, 30 V, and 40 V.
185
2MS/s
TekE
g 20MS/s
AcqsiTek
2161
-.
M
M
fit~
~
-.
..
-i
-i
-
11i
191 Acqs
....--..
2-'
Tek I
g20MS/s
=
Aug 191
14
14:45:1
243 Acqs
Cnz
105MVN
M
20V
TekEEN 20MS/s
Ch2
2.59S
1.
172 Acqs
20.K V
14Au
Averages: 256
Ac
+
.. . . . . . . . . . . ..
0
le
t
[on
......
......
JTh
......
Sample
U
2
Peak Detect
SMS/s)
q
. .............
.............
.
q I
......... ....
J ... ...
... ............
....
Envelope
9
t
7
2-s
M
46
2
..... mVI
4
20V4 W M2.
h ! 2
X
20.9V
4.
-,i'
o fter
ItS utoni
Aug 6i.1
15:20
14
Fig. 8-18: AC output voltage ripple (CH) for RF output powers of 1 W, 20 W, 30 W, and 60 W. The
output ripple is 1.8% of the DC voltage at 1 W output but only 0.8% at over 20 W.
lek
U 20MS/s
ThkUEU 20MS/s
182 Acqs
404 Acqs
(b.
(a)
-44
4
1+
-..-..-..
1.2
h~
.h
.
.
....
4
-..-
Tk
U
Lni
1000
11?:
4 Aug
12: 04
540 Acqs
20MS/s
Idu v
*mKa-sssmrv
+-...-
m M2.5sJ
cii
-,
ino v
log
14:Au%
17:
10:20
T-
(c)j
14
M.
W ..........
IN
Chi
2eV
flu sogmv
C
Mzsps Chi
I.
23.6
Fig. 8-19: VSR voltage output (CH1) and filter inductor AC current (2 mA/mV, CH2) for RF output
powers of (a) 1 W, (b) 20 W, and (c) 60 W.
186
Fig. 8-20 shows a plot of the center-tap AC current and the voltage at the drain of
Notice that the current is
one of the switching FETs for the DC-to-RF converter.
relatively flat but has large switching spikes approximately one-tenth the size of the DC
current.
TekMKoU SO0MS/s
881 Acqs
..
..
..
..
..
.
....
..
...
........
.........
...
....
...
..
7
1 -+
..... . . . . . . . . . . . . . . . . . . . . . ....
. .. . . .
. . . . . . . . . . . . . . . . .
..... . . . . . . . . . . . . . . . . . . . . . ....
Chl
20 V
L2
. . . . . . . . . . . . . . . . . . . . . ....
20mV
. .. . . . .. . . .
M loans
. . . . . . . ....
. . . .
Ext
. . . . . . . . . ....
X
1.5 V
14 Aug 1998
16:36:29
Fig. 8-20: FET drain voltage (CH) and AC current into the (2mA/mV, CH2) at an output RF power
of 10 W. The DC current is estimated to be 0.9 A. The current spikes are on the order of 80 mA.
8.4.1.4 RF Filtering
8.4.1.4.1 Topology
Fig. 8-21 is a schematic of the filter for the RF output of the transformer. In the
original design plan, a 5-pole Chebyshev 50 ft to 50 ft filter was to be implemented.
However, the filter did not work very well and was replaced on the desktop by a 4 pole
187
low pass filter that converts the 50 fl magnitude load to a higher magnitude impedance at
the transformer. A plot of the voltage transfer function is found in Fig. 8-22.
L13
L14
To Load
s
T1
0
10
turs
C5 4p
. ,..
'5 3.
2400pF
..
18 00pF
5
turns
Fig. 8-21: Schematic of filter for Channel 4 on ultrasound driving card.
Fig. 8-22: Simulated voltage transfer function of filter.
8.4.1.4.2 Components
The filter capacitors are through hole dipped mica capacitors (high
Q)
with a
voltage rating of 500 V. The inductors are iron powder toroidal cores wrapped with 24
AWG magnet wire.
188
8.4.1.4.3 Waveforms
Fig. 8-23 shows a typical input and output voltage waveform for the filter
powered at 60 W.
3588 Acqs
SOOMS/s
TeklWUO
..........................
Mu
....
....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2-+
....
.
EWJ
.
................................
.......... ........... .......... .......................
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.. .
.
...... .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
100 V
C12
.
.
.
.
.
....
100 V
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. ...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. ...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. ...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
....
.
M 100ns
Ext
.
~X-
1.5 V
14 Aug 1998
16:15:1 9
Fig. 8-23: Input (CH) and output (CH2) voltage waveforms for 60 W of RF power delivered to a 50
11 load.
8.4.1.5 Power Coupler/Diode Detector
8.4.1.5.1 Topology
The RF power is measured by using a dual directional coupler and a compensated
diode detector (Grebenkemper 1988; Grebenkemper 1990). A circuit diagram of the dual
directional coupler is found in Fig. 8-24. Assuming that the transformers are ideal with
189
turns ratio of k (18 in the actual system) and setting the coupler standard impedance R to
50 Q, then the function of power delivered to the load PL may be written as
k2
1
V) R V +(I+ k 2 )V*]
PL =(Vf
where Vf is the forward power signal and Vr is the reflected power signal (* represents
complex conjugation). These voltages in turn are determined by the load impedance ZL
and driving voltage VD as written in the following equations:
I Z
2
(1+
Vf =-VD1
-+(
k
1
Z
+
Z
2
k3 + k
+2
k)
R
Z
V,.=-VD 1
(1 - -)
R
1 2
Sk3
k
Z
R
Notice that when Z is matched as equal to R and k is large, then Vr is zero and
1
Vf =-- VDk
PL
fVf R'
Since the maximum rated power for the ultrasound driving system was 60 W into a 50 f
load, the peak voltage at the load is 78 V.
The turns ratio was chosen as 18 to keep Vf
under a 5 V magnitude while not saturating the toroid or causing significant insertion
loss.
It is important to note that the voltages called "forward power" (Vf) and "reflected
power" (Vr) are proportional to the voltage magnitude presented to the load and scaled by
factors comparing the mismatched between the load circuitry and the 50 fl ideal case.
190
Therefore, a measurement of "reflected power" power during an ultrasound sonication
mainly represents an electronic load mismatch and not necessarily ultrasonic waves
which have been transmitted, reflected off of an acoustic barrier, and received by the
ultrasound transducer. While this condition can and does occur in some circumstances, it
is more likely that a high reflected power indicates a faulty cable or poor electronic
matching for an ultrasound channel. The degree of mismatch will regulate the amount of
actual ultrasonic power transmitted (see Chapter 2 for an acoustic plot).
Dual Directional Coupler
................................................
Output of
Amplifier
T1
Load Impedance Z
Matching
n
1
k
Transducer
T2-=
R
IR
Fig. 8-24: Schematic of the dual directional coupler.
A schematic of the tandom match diode detector (Perras 979; Spaulding 1984;
Grebenkemper 1988; American Radio Relay League 1989; Grebenkemper 1990) is found
in Fig. 8-25. The simple detector consists of a diode connected to a low pass filter with
cut off of 160 Hz (or a decay time constant of 6.3 ms). The tandem match op amp circuit
compensates for the diode drop at low power levels. This accomplished by the matched
diode (D29, D32) in the feedback loop and the compensation resistor (R38 and R41).
The value of the compensation resistor is smaller than the resistance of the low pass filter.
This is because the voltage across the low pass filter resistor represents the average
current flowing through the source diode (D27 or D28). The peak current through the
191
source diodes (at the peak voltage) can be significantly higher than the average current.
Therefore, the feedback compensating resistors (R38 and R41) are smaller than the low
pass filter resistors to increase the current through (and voltage across) the feedback
diodes for a more accurate compensation at low power levels.
Vf
Vr
D28
DL4148
D27
DL4148
lOOK
R38
]O.O01F
LMC6484
DD32
2K D148
U53B
R43
lOOK
_-C64
C.6F LMC6484
Fig. 8-25: Compensated diode detector circuitry.
8.4.1.5.2 Components
The dual directional cores are 0.5"diameter ferrite toroids (material 77, Amidon)
wrapped with 24 AWG magnet wire.
The 77 material was chosen for its high
permeability (about 2000 up to 1.8 MHz) and high saturation flux density (4000 gauss,
estimated flux density was 700 gauss for 70 W into a 50 fO load.
The diodes were ultrafast switching Schottky diodes.
8.4.1.5.3 Waveforms
Fig. 8-26 shows the output voltage of the RF amplifier and its sensed voltage by
the dual directional coupler for a 50 fk load at 1 W.
192
TekM
2997 Acqs
SOO500Ms/s
T.................................
00M...../s....2997.....Acqs...
..
AI
1 -+
Chl
500mV
M
10 V
M lOOns
Ext
200mV
X
14 Aug 1998
13:55:25
Fig. 8-26: Voltage wave forms of output voltage (Ch2) and dual directional voltage (Chl) for a 1 W
power into a 50 il load.
Tek Run: 500MS/s Sample
lADMd
.......
T
........................
... .......
............................
....
.........................
...
...........................
....
.........................
. . .
. . . .
. . .
.... . . . . . . . . . . . . . . . . . . .
.... . . . . . . . . . . . . . . . . . . . . . . ...
PAm ! 11 11
Lfjq 101,111,11.4.111,1110,
.111
.
. . . . . . . . . . . . . . . . . . . . . . . ....
....
. . . . . . . . . . . . . . . . . . . . . . ...
2-+ ...
am
5;0,0,, V,
M 100ns; Txl'
...2;0'0"i V-
14 Aug 1998
14:21:47
Fig. 8-27: Input (Chl) and output (Ch2) of the op amp compensation circuit for a 1 W load.
193
8.4.1.6 Transient Response of Power Loop
A block diagram of the model of a the transient feedback loop is found in Fig. 828.
The power loop contains two feedback paths. First, a low frequency path flows
through the power measurement components back to the integrator.
Second, a high
frequency path exist through the frequency compensation R-C circuit. Most of the block
transfer functions are straightforward to compute with the exception of the DC-to-RF
converter. This was measured empirically instead of being modeled (see Fig. 8-29 for
equations and values). Fig. 8-30 contains a transient response plot of the output voltage
at turn on to 60 W with a fixed DC voltage supply (power loop disabled) which was then
modeled as a single pole response. Fig. 8-31 shows the simulated DC response time of a
step input indicating a stable response time of about 40 ms. The response time with the
power feedback was a bit faster, measured to be 10 ms (see Fig. 8-32). The difference
between these measurements is most likely due to the simple modeling of the VSR
internal response or the simple model of the DC-to-RF converter.
Input Gain
Power Set
-
-
(s)
Integrator
H
s(s)
VSR
VSR Filter
DC-to-RF converter
Ho s)
Hi(s)
H,(s)
Point
E
Freq. Compensation
H,(s)
Power Feedback
H
-jI"(s)
H
Fig. 8-28: Block diagram
Amplitude Detector
Ha, (s)
of power feedback loop for transidnet analysis.
194
Output
sRde ResrCj + Rde
2
s Lf (Rde + Rsr)Cf + s(RdeReCf + Lf + RRdeCf + RResrC) + Rde +
1e6
+5e5
) =
)s
where:
sCyp
C 4 is the frequency compensation capacitor(0.1 uF)
+1
S
Hv(s)=-
RvjsC, + 1
(s)
=
+ sCint
+
Pb
Rv
1
+
1
Hi.,(s)
RvIb is thefrequency compensation resistor (100 ohms)
Rde
is the effective DC resistanceof the Class D/E stage (20 ohms)
Cf is the VSR filter capacitor(220 uF)
sC
Lf is the VSR filter inductor (47 uH)
Resr is
the equivalent series resistanceof Cf (0.1 ohm)
Hvsr (s) = Gvsr
R, is the DC resistanceofLf (0.02 ohm)
Hd,,( s) = -/n
RIb is the powerfeedback resistanceto the integrator(10k ohm)
Cnt is the integratingcapacitor (0.15 uF)
Hpjb(S)
-p
Gvsr is the voltage gain of the VSR compared to inputpin Vc (15)
n is the turns ratio of the power meter (18)
Fig. 8-29: Transfer function equations for transient analysis.
Tek Run: 5OMS/s
Sample
r
-r
ErM
I1
L
A.: 179 V
@: 92 V
...
. . . . . . . . . . . . . . . . . . . . . . . ....
...
. . . . . . . . . . . . . . . . . . . . . . . ....
.... . . . . . . . . . . . . . . . . . . . . . . ... . .
2-
. . . . . . . . . . . .
. . . .
. . . . . . .
. . . .
.
....
...
....
....
......
----------------------------... . . . . . . . . . . . . . .
. . . . . . . ....
...
...........................
...............
2-*
. . . . . . . . . . . . . . . . . . . . . . ....
...
.........................
..........................
....
MjJ
so V
Ch2
5 V
M
l Is Ch2
-.
4.2 V
15 Aug 1998
08:55:39
Fig. 8-30: Transient turn on response to 60 W output power at 1.5 MHz using a fixed DC supply
voltage.
195
Step Response
18
16
14
12
10
-0
8
E
6
4
2
0
005
0
0.15
0.1
0.2
Time (sec.)
Fig. 8-31: Power feedback response from a step input change.
Tek Run: 1OOMS/s
Sample
N-
I
Delay Time: 10.0007325ms
A:16I
A:: 162 V
Time Base
@(:. -77 V
Main Only
WF
Intensified
Delayed Only
. .. ... .. ..-...
..
-..
..
-..
2-+
+Delayed Runs
10.0007325m
millI
30 V
ime Bse
Trig
Ch 2
igger
Position
256
M 2.5ys
D 500ns
Ch2
472 V 1
Set to Min
Fit to
Screen
off
Fig. 8-32: Transient reponse time to 60 W output at 1.5 MHz using the power feedback
196
loop.
8.4.2 Phasing Loop
8.4.2.1 Overview
As described in Chapter 2, the object of the phasing loop is to regulate the phase
of an individual channel such that its output electrical sinusoid is in proper phase as
compared with the other array elements.
The loop consists of three sections:
phase
generation, phase correction, and phase detection. The timing circuitry to coordinate all
of the channels is done using the fast edge, AC logic ICs and a synchronized control
signal (Step Stack) and the distributed high frequency clock (running at 16 times the
ultrasonic frequency).
8.4.2.2 Phase shifting
8.4.2.2.1 Method
The input clock signal to the ultrasound driving card is operating at a frequency 16
times the ultrasonic frequency. As described in Chapter 2, this input clock is put through
a modulo- 16 binary counter with a preloaded offset such that the divide-by-8 output is a
0-5 V square wave of the desired ultrasound frequency with a multiple of a 22.5' phase
shift. This signal is then put through a programmable delay chip to gain the fine phase
resolution.
8.4.2.2.2 Components
The counter chip is the 74AC163.
This chip has a maximum frequency of 90
MHz. The fine delay is gained through the Dallas 1021 8-bit programmable delay line
with 0.5 ns increments.
197
8.4.2.3 Phase detector
8.4.2.3.1 Method
Class D/E amplifiers are not linear, and therefore their output phase will change
depending on power level and loading.
This variable phase is unfortunate for arrays
whose elements are different sizes and impedences. For this reason, the output phase
must be detected and regulated to guarantee the desired electronic phase into the
transducer. The output phase is regulated through one of three feedback signals which
are multiplexed into the correction circuitry.
The first feedback signal is the default
disabled feedback signal. This signal originates directly from the phase shifting circuitry
and does not require any conversion back to a 0-5 V square wave. The second type of
feedback signal is sensed between the amplifier output voltage and the matching circuitry.
This is called amplifier feedback and it passes through a detector circuit to convert the
high voltage sinusoid to a 0-5V square wave. The last type of feedback comes from a
source external to the ultrasound driving card, generally from between the transducer
matching circuitry and the transducer. This is also a high voltage sinusoid that must be
detected and converted to a 0-5 V square wave. It is called transducer feedback.
Fig. 8-33 contains the detection circuitry for channel 3 of the ultrasound driving
card. On the left are the two high voltage sinusoid inputs PF3 (transducer feedback) and
RFOut3 (amplifier feedback) and the feedback disabled option is represented by
RFClock3.
The inputs of the multiplexer (74ACT253) determine which of these
feedback signals are input to the phase correction circuitry. The detection circuitry for the
amplifier and transducer feedback are almost identical. They consist of a circuit using a
198
high impedance input resistance (5kfl) grounded through antiparallel diodes into a
comparator chip. This was used to convert the voltage sinusoid of up to 200 V p-p down
to a t 0.5 V square wave. A simple resistor divider was not used since the small
amplitude sinusoids would yield very small input signals which can approach the noise
level of the card. The antiparallel diodes force the voltage to be above noise.
+5V
U51
R117
5
T0
MD353
IOOK
DCO
0
U49A
210
Y
C
42
R119
10K
D58A
W
Fig.
5
D58B
E
C
ifd 2
&e 5
8
MAX901
8-33: Phase detection circuitry.
8.4.2.3.2 Components
Again, AC or ACT logic is used for these circuits to avoid dull edges.
The
antiparallel diodes are schottky barrier diodes with low capacitance (10 pF) and fast
reverse recovery time (5ns) such that they do not cause undesirable phase shifts in diode
charging time. Maxim's MAX9O1 comparator is used to convert the signals to digital
square waves.
8.4.2.4 Phase correction
8.4.2.4.1 Method
The necessary phase correction error signal such that the feedback signal and the
set point phase are coincident is accomplished by using a phase locked loop (PLL). A
schematic of the phase correction circuitry for channel 3 is found in Fig. 8-34. The PLL
199
used is the Philips 74HCT9046.
Basically, the PLL compares two input signals and
adjusts a voltage controlled oscillator according to the relative phases of the inputs. If the
output of the oscillator controls one of the input signals, the oscillator frequency and
phase can be "locked" such that the input signals are synchronously phased.
In this applications the input signals to this circuit include the desired phase signal
from the phase shifters (Phase Clock 3 to SigIn) and the phase feedback signal (to
Comp). The PLL uses an internal phase comparators which has a 3-state output current
pump to either source, sink, or shutoff current through pin PC2. This current is pumped
into a damped integrating capacitor (R66 and C147) to produce the low pass DC voltage
for the internal voltage controlled oscillator (Vcojin).
The output of the voltage
controlled oscillator (Vco-o) is then fed through a gating circuitry and the input of the
DC-to-RF converter. The stability of this loop is controlled by the additional circuitry of
the PLL and was determined using the 9046 data sheets. First, the frequency locking
range was chosen using a center frequency of 1.5 MHz and a locking range of t 0.4
MHz. Using these requirements the PLL values of RI (R70), R2 (R69), and C (C145)
were determined. Second, the output of the comparator was low pass filtered using a
passive filter with damping (R66 and C147). This filter crossed the unity gain around 10
KHz. The magnitude of the output current pump (or sink) was determined by the Rb
resistor (R68).
The other circuitry connected to the PLL is to monitor for proper operation or to
instigate a safety shutdown. For example, the circuitry connecting the input phase clock
signal and the INH (inhibit high) pin is designed to shut down the PLL if there is a loss of
input clock signal. This prevents the oscillator from randomly ringing in the case of input
200
loss. The output of another phase comparator (PCI, an exclusive-or comparator) is used
through a low pass filter as an indicator that the output signal is locked (this is then fed to
the microcontroller to cause a software shutdown if the PLL does not lock in a specified
time). Lastly, the output of Demout mirrors the input voltage to the VCO such that the
output frequency of the PLL can be monitored by the microcontroller.
+5V
R92
1K
D51 DL4148
7
R95
I
D63
DL4148
5
C156
P.asLf
Tn45
Fig.
pF
ni 8
R129
U50
N
PCI
R66
CnBsoe
abckdarm
c4
12
R69
Phase Feedback Signal the
wsod
R
Comp
ci
o
O.IpF
9
10
RK
i
abu
2
1010
IK\ ECO: Add Component
Fig. 8-34: Phase correction circuitry using the 74HCT9046 PLL.
8.4.2.5 Transient Response of Phase Loop
Fig. 8-35 contains a block diagram of the phase correction circuitry. The gain
blocks are determined through the PLL's external circuitry as described in the part
specification sheet. The DC-to-RF converter and the phase detector were modeled in a
first order as constant unity gains. The simulated results yielded a locking time of about
400 Ls. Experimentally the phase was found to lock in about 200
201
ss.
Voltage Controlled
Oscillator
Low Pass
Filter
Phase
Cornarator
Phase Shifted
Clock
DC-to-RF
Plant
o RF Sinusoid
KDE
- K,cO
KK
Phase Detector
Kde
Where:
5
R 4=Lowpassfilterresistor (1000 ohm)
4 x )T
I+ sR 4C2
/ A +sR 4C2 I /17
=
=
K
C2 =Low passfilter capacitor (0. 1uF)
fL=Lock band (0.4 MHz)
A =Gainof internal erroramplifier (105)
2 fL x 2)7
Vf~ - 1.1) - 1.1
V, =DC supply voltage (5 V)
Fig. 8-35: Block diagram for transient analysis of phase correction.
Step Response
-...
0.1
.
..
..
.
..
..
.
..
..
.
..
..
.
..
..
.
E
0
1
2
3
Time (sec.)
Fig. 8-36: Simulated transient locking response of PLL.
202
4
5
6
x 1T4
9. APPENDIX B: PARTS LIST AND SCHEMATICS
9.1 Parts List
QTY
4
4
8
16
Part No.
415-0932
2843010302
FT50-77
10KH-ND
2 74AC04SC-ND
1 74AC08SC-ND
4 74AC163SC-ND
1 74AC32SC-ND
1 74AC541SC-ND
I
1 74AC74SC-ND
Designator
L3,L7,L1 1,L15
T1,T4,T7,T10
T2,T3,T5,T6,T8,T9,Tl 1,T12
R109,R110,R1l1,R112,R113,R114,R11
5,R116,R117,R118,R119,R120,R121,R
122,R123,
R124
U2,U44
U26
Description
Inductor 47pH
Dual Apperature Core
0.5" Torroidal Core
IOK OHM 1/2W 5% CARBON FILM
RES
U32,U34,U40,U42
U14
U13
IC 4-BIT BIN CTR SYN RE SO16
IC QUAD 2-INPUT OR GATE SO14
IC FACT OCTAL BUFF LINE DR
SO-20
IC DUAL D FLIP FLOP TRI-ST SO-
U43
IC HEX INVERTER SO-14
IC QUAD 2-INPUT AND GATE
S014
14
2 74ACT253SC-ND U47,U51
IC DUAL 4-INPUT MULTIPLEX
1 766-163-R1K-ND
RP1
RES-NET 1K OHM 16DIP 8RES
1 766-163-R4.7KND
8 BCX70KCT-ND
RP2
1 H2131-ND
J2
SO16
SMD
Q3,Q6,Q9,Q12,Q13,Q14,Q15,Q16
8 IRF634S-ND
40
1
4
4
1
Q1,Q2,Q4,Q5,Q7,Q8,Q1O,
Ql1
LL4148CT-ND
D1,D2,D3,D4,D5,D7,D8,D9,D10,D1 1,
D12,D13,D15,D16,D17,D18,D19,D20,
D21,D23,D24,D25,D26,D27,D28,D29,
D3 1,D32,D49,D50,D5 1,D52,
D61,D62,D63,D64,D65,D66, D67,D68
LM4040AIM-5.0- U24
ND
LMC6484AIMU4,U7,U10,U53
ND
LT1074HVCT-ND U3,U6,U9,U12
1_
MAX51OBCWE- U25
RES-NET 4.7K OHM 16DIP 8RES
SMD
TRANS NPN 45V SMD SOT23
HEADER 2MM 6 POS SMT
RT/ANGLE
HEXFET 250V 8.1A N-CHAN
SMD220
ULTRA-FAST SWITCHING DIODE
S/MT.
IC PREC VOLT MICROPWR 5V REF
S08
IC CMOS QUAD OP AMPLIFIER
S014
IC 5A STP DN SWITCH REG T0220-
5
IC QUAD D/A SERIAL 8BIT SO16
ND
1 MM74HCOOMND
1 MM74HC138M-
U21
IC DUAL 2 IN NAND GATE SO14
U23
IC 3-8 LINE DECODER S016
203
ND
1 MM74HC165M-
U22
IC PAR IN SERIAL OUT SO16
U29,U30,U31,U33,U37,U38,
IC 8 BIT SHIFT REGULATOR SO16
ND
8 MM74HC595MND
25 P1.00KFCT-ND
U39,U41
R1,R2,R12,R13,R23,R24,R34,R35,R52,
R54,R59,R66,R7 1,R76,R79,R84,R85,R
92,R95,R100,R1O1,R127,R128,R129,R
RES 1.00K OHM 1/8W 1% 1206 SMD
130
4 P1.00MFCT-N
R7,R18,R29,R40
RES 1.00M OHM 1/8W 1% 1206
1 P1.62KFCT-ND
R46
SMD
RES 1.62K OHM 1/8W 1%
1206 SMD
20 P10.0KFCT-ND
RI 1,R22,R33,R44,R47,R48,R49,R57,R RES 10.0K OHM 1/8W 1% 1206 SMD
58,R61,R63,R69,R70,R73,R108,R125,R
24 P100FCT-ND
R9,R20,R31,R42,R131,R136,R137,R13
132,R133,R134,R135
RES 100 OHM 1/8W 1% 1206 SMD
8,R3,R4,R14,R15,R25,R26,R36,R37
12 P100KFCT-ND
R6,R1O,R17,R21,R28,R32,R39,R43,R8
RES 100K OHM 1/8W 1% 1206 SMD
0,R86,R96,R102
8 P12.1KFCT-ND
R5,R8,R16,R19,R27,R30,R38,R41
RES 12.1K OHM 1/8W 1% 1206 SMD
1 P2.2MECT-ND
4 P24.3KFCT-ND
4 P365KFCT-ND
R50
R83,R88,R99,R104
R56,R60,R68,R72
RES 2.2M OHM 1/8W 5% 1206 SMD
RES 24.3K OHM 1/8W 1% 1206 SMD
RES 365K OHM 1/8W 1% 1206 SMD
2 P4.75KFCT-ND
R126,R45
RES 4.75K OHM 1/8W 1% 1206 SMD
3 P5637-ND
C65,C66,C68
330UF 10V HFQ ALUM RADIAL
8 P5766-ND
C9,C10,C25,C26,C41,C42,
220UF 50V HFQ ALUM RADIAL
C57,C58
CAP
C70,C71
22PF 50V CERAMIC CAP 0805 SMD
C137,C145,C148,C158
560PF 50V CERAMIC CAP 0805
C1,C17,C33,C49,C73,C75,C77,C79,C8
1,C83,C85,C87,C89,C91,C95,C97,C99,
CAP 3.3UF IV TANT TE SERIES
CAP
2 PCC220CNCTND
4 PCC561BNCTND
23 PCS2335CT-ND
SMD
C101,C103,C105,C107,C109,C160
4 SK36DICT-ND
D6,D14,D22,D30
DIODE 60V 3A SCHOTTKY SMT
1 X428-ND
Xl
12.000 MHZ QTZ CRYSTAL(HC-49)
2
4
2
8
U18,U19
L4,L8,L12,L16
U45,U49
C4,C8,C20,C24,C36,C40,C52,C56
EEPROM
Inductor 22pH SM
IC QUAD COMPARATOR SO-14
CAP 200pF 500V Mica SM
CAT28C64BJ
IHSM-5832 22pfH
MAX901BCSE
MC18FD201J
4 74HCT9046A
U46,U48,U50,U52
PLL
69 140-CC501B103K C2,C 1,C 12,C 13,C 1 6,C 1 8,C27,C28,C2 CAP 0.01pjF 50V Ceramic 0805
9,C32,C34,C43,C44,C45,C48,C50,C59,
C60,C6 1,C64,C72,C74,C76,C78,C80,C
82,C84,C86,C88,C90,C94,C96,C98,C10
0,C102,C104,C106,C108,C1 10,C1 1 1,C
112,C I13,C1 14,Cl 15,C1 17,C I18,C 119,
C120,C121,C122,C123,C124,C125,C12
6,C127,
204
140-CC501B103K C128,C129,C130,C131,C132,C133,C134,
C135,C159,C161,C162,C163,C164,C165
21 140CC502Z104M C3,C7,C15,C19,C23,C31,C35,
C39,C47,C5 1,C55,C63,C69,
C139,C141,C147,C149,C154,
C155,C156,C157
4
4
4
8
4
2
4
1
140CC502Z104M
CDV19-FF182J03
CDV19-FF242J03
PT1060-E
DS1021S-50
MC34064
MC34152
MC68HC11KOCF
N3
8 MMBD353LT1
CAP 0.1pF 50V Ceramic 1206
C14,C30,C46,C62
C6,C22,C38,C54
C5,C21,C37,C53
L1,L5,L9,L13,L2,L6,L10,L14
U27,U28,U35,U36
U16,U15
U1,U5,U8,U11
U17
CAP 0.15 use 0.1 pF 50V
CAP 1800pF 500V Mica
CAP 2400pF 500V Mica
Inductor Torroid Cores
8-bit delay
IC Undervoltage Sense
FET Drivers
Microcontroller
D53,D54,D55,D56,D57,D58,D59,D60
SS DIODE SOT23 T/R3K
9.2 Schematics
205
A
a
0
6
__________
Ji
1
RFout
RFout
PFI
PF2
4
3
5
7
5
11
+7V
+48VC
+48V
MISO1
i.MOSI JSCK
J_SS -JEnable
JRecord
Power
JStepStack
JCard Reset
-J 161
+5V
-5V
+7VC
-
1
1
23
7
29
31
33
57
41
45
47
3
51
PF4
Phase
Feedback
53
PF4
RFout3
RFout
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
01
53
55
2
. 2
4
6
6
10
12
12 14
14 1
J
19
20
22
24
26
28
30
32
34
36
36
40
42
44
46
48
50
52
54
56A
SS
6
JRoRecord Power
JAStepStec_
16
0+48V
JCard
1
4
J-Enable M
- +7V
1
23
JMOSI
JSCK
Rse
J-161
A1
A2
A3
A4
Y
Y2
AS
A7
Y6
AS
Ye
Y3
Y4
8
17
'mos
4
SCK
ss
1S
Enable
Stp Stck
Card Rset
SG1
2
24
CA6
CAS
-JCA5
CA4
CA3
CA2
CAI
270
32
34
36
G2
74AC541
CAO
.- 1le
+5V
40
3--
-5V
42
44
46
+7V
JLSCI RxDG
4
6
52
SCI RoD
3
+5V
74HC32
56.
R126
4.75K
CON56A
SCt TxD
10
JSCI TeD
74HC32
+5+
C65
330pF
10V
+,V
0
J2
I
J_-SCI
J-SCI RoD
4
MODE SELA
MODE SELB
3
+7V
TxD
2
C66
10V
330pF
CONS
-5V
C66
1330pF
110V
Brigham and Women's Hospital
Department of Radiology/MRI
LMRC, Room 007
221 Longwood Ave.
Boston, MA 02115
Input Section
Size Iocument
B
(Doc)
a
I
I
D
)a
Number
C
r
n
1995
Sheol
0o
o
Aa
a
+5V
SPI Address Table
RP2
4.7K
0x00
NOTE:
Upon startup. the SPI address
ducoder must be written to with a
vozero value to load ihe cord
into the shift register.
1
U22?
10
11SE
CAO
CAI
-
-a--res
--
12
14
CA2
OxO
0x02
0x03
0x04
Co05
C
0 xO
0x07
Card Address Shift Register
Power ControlDAC
Unused
Unused
Phase Shift I
Phase Shilt 2
Phase Shift 3
Phase Shift 4
3 D
CA3
CA4
4E
1
CAS
CA
1
6 a
CA7M
OH
H
15
LK
INHC
9
GSDO
--X
H 7
74HC165
3
U2F
74AC04
+5V
46
SPI
y
Y
YI1 1
A5Y
SPI
1~
13
3C
C
Gt
G2A
_EAG2
(3y4
YYG4
Y5
Ys
Y7
10
Pssell'
Ph S912'
Ph:Sel3'
E
P
hsS*14'
74HC138
I,
+-7V
0)
N'.
R52
1.00K
- --= Vref
+5V
U2?5
RESET=
LOAD
So l
SCLK
14
j
10
EUI
CS
SDI
SDO
CK
DGND
VSS
F
U24
LM4040-5
4
REFCDt
OUTA
OUTB
OUTC
OUTD
Power Set
CH2 Power Sal
CH3 Power St
CH4 Power Set
CHI
AGND5
'
Brigham and Women s Hospital
Department of Radiology/MRI
LMRC, Room 007
221 Longwood Ave.
Boston, MA 02115
Serial Peripheral Devices
549
a
11
)aIe:
0
Documasn Nomber
IV
(Dcc}
hursday
November
C
30 1995
hoat
10 of
12
V
A45
4.79K
V MC34064
P46
1.62K
U44F
0.IpFcog
74AC04
+5V
+V
+5V
+V
Ule
'47
48
-
10.5K o.OK
R125
10.0K
MODE MO D.
MC34064
10.0K
49
OD~TeY
MOD SK.
I
XOUT
-
__
?W VppO
4k
Vrf
VAl
4L
I
V.f4 t
1.
R-PAC.
t
D3
D4 t
AS
05
Do
D
Flw
0
AP C
I
1
DATA0_PCo
DATASPCI
DATA2_PC2
DATA3PCS
DATASPC
DATASPC"
DATA7 PC,
XA13-P 0
XAIO4PGGI
XAIIPO2I
XA IPGS3
AVVe
PL;3
EaPL 4
XA I7_PG4
XAtSPoll
00
7HC
Pert PHO
2
PW.
rP.
PW4-PH3
MSPD,
MOS
01
AG
A
- -E
ADDR14_P9S
ADDRIS..P7
72 RDPDO
T..DPD
UI.SOCPDZ
I
A4
A12
ADDRI2_P94
PASIC3
PAIJ C2
PAD:C I
IOCI
PASIC4.
PA4..C4_.OCI
PAS-OC3CI
0A6OC2OCl
PA7_OCIPAI
Enabe
ILtPB3
ADDRI2_PB4
PEZAN?
step Sscak
O
A
IDP52
ADD
ADOR
At
A7
ADDRI.PB
ADDARSP i
PESANO
PEIANI
PE2_ A.2
PE3_AN3
PE4_AN4
PESANS
Vreft2
I
ADD.2.PF1
ADDR3_PF3
ADDR4_PF4
ADDR._PFS
ADDR
ADDRr__PF6
PF7
AVdd
AV..
RPI
AVVe
ADDRIPFI
z a orlef
svr[EZL
1:18
MOD..ADDR..PFS
PDI
PWP
write
MODE SEL
PH
2
-
Protect
At
c
As
A.
00
C0
IN
D
D
4A
D
A?
A
D4
02
CSPt-PH
Sc,~
.DeJ
VdPV
Ves
ExTAL.
L
-.
Yd
FMfts88
XTAL e6__
U44E
74AC04
SMC6SHCIIK0CFN3
SPI2
1 5D0
MIS;
12.0
act
aCt
P.1
7
Addr...
LI...
MHz
2.2M
I
I
2pF
'F
-- "-""]LOAD)
> Shutdovrnl'
L ----- **'U448
74MC00
14A.
---= hutdoen2'
U44C
74HCOO
. I
I ! II I 'll> 0 - -<
heido-ns
4o
74HC06
Brigham and Women's Hospital
De.,7.
ra07f
MORC. Room
13 !?
74
AC104
hudw
1w
7414C00
f""
CS-
-
I
a.boo:nday
lfgmc6
1De
CI
diokWg/MF"
9014=1111II
MIe-fcetreo.r
R
.#f DcmeNo.".r
Noyomw 06. Ive
sh"I . 10
12
I
OUT
IN
---
SEA
--
1- ---
CLK
CA
--
-I
-
OF
s
a"
P,
74
-- ,
H'I
SEA
-'1--
4Po"O'
-
a
oo OW
P=o
SEA
CA
RCLK
OCC
SRCLR 00
RCLK OF
a
00 ---
-
3
IIA
-
RCLK
74HCSO
O
ENP
NTK
OUT
E
SA~03
QPh"*e Clocks
Pa/0
CA
o
DIN
P3 D
D1IPSI.0oT
SACLA
.
P
0'
00
a's
8
-
f
-81021,1o
S
164CS
0E
OF
CLK'
IRL
OF
OH
OF*
4
.,v
-,--2-.
74MCSOS
SE
DC 9 -tS
-o
-t-1I
I
Phasel4"
G 0N
74"C69S
OUT
SA
--
.
CLK
CIOckS
RO
SEA
I
,
CP
+5v
RCLK
j
gi
c
IN
-
.
74ACJ63O
F
C
.
A
a
0
~
SE
..
P-0,0
.5
IsA
A
OUT 1
IN
P2
-s DS10218SO0
74HCSOS
-
I
E
-
'-''
'
CA
S -ICC
CW
NCO
I-IA
.
A
0.3
CLA
CLA
74ACIS30
I
SDI tDtl
gotc
toAo
___________
JI
___________
J
,Sv
.5
UZE
-
CLE
744AACC7
4
Brigham and Women's Hospital
DepIUYt
OCd.A
,M0
of
80.1" A*~ 115
C
(Doot
M*
I
C
A
a
I
+5V
+5V
R76
R109
10K
1.00K
Rto
00K
D49
U45A
DL4148
MMO0353
PF1
013
OCX 70
102
R79
1.00K
0530
10K
D53A
2+
Hill
10K
MAX901
-
061
U46
DL4140
INH
C37-
U458
D548
R112
10K
c
560pF
C154
11
+
MME>353
RFoutl
T
R12 7
PCI
PC2
IA
677 I
Rb
2
13
1.0
15
.00K
R56
Phase Clocki
0.IpF
14
R57
C139
-=.
R2
10.0K
365K
Sig_,n
12
MAX901
TCr62
0.01 pt
R04
4-
5
054A
PLit
OK
-
0
Vco-in
R58
R83
Demout
Ri1
24.3K -
10.lOK
Camp
ShHdown
Vco-o
3
RF CLOCKI
74AC00
74HCT9046A
R132
RF
I CO
CLOCKE:
y
7
2Y
2
9
1Cr
1C2
tC3
2C
112C0
12 2Cr
RF CLOCK
F 84
14
1IG
Shutdown2-
DL4148
U48
ECO: tlfrpts ISS
end add jumpers
l00K
055A
PC1
CIA
PC2
7
Rb
2R126
//
13
15
JPL12
y
1.00K
l
C1K
R59
1.00K
0.0l
'' I
C0
*385K
Sig_ln
Phass Clock2
12
R2
C141
=-0.1pVF
Veoan
0
R88
Demout
RK3
10.
24.3
RI
3
9
INH
R6
DL4148
=
12
.
-
4
4co
Camp
HFIF CLOCK2
Shutdown2'
MAX901
R133
I-
.
13
056B
D56A
10
74AC05
74HCT9046A
U450
MMBD353
10K
C158
560pF
O.1pF
U45C
0558
IIS10K
RFout2
C155
--.
KG
0 +
5
R85,
1.00K
ROs
5
MM0303
R114
10K
BX70
2G
74ACT253
R5V
HI 13
10K
PF2
A
2
ECO: Add Component
D50
2C3
PFA
PFA1
Shutdownl
----A-^.*%10.k\
+5V
_
ECO: Add Component
/
10.0K
Brigham and Women's Hospital
Department or Radiology/MRI
LMAC. Room 007
221 Longwood Ave.
Boston. MA 02115
MAX901
Phase Correetion. Ch
ze
0
ale:
.
1& 2
Document Number
(Doc)
Saturday,
December 02, 1995
C
Sheet
E
2
o
12
ID
a
IE
I
+5V
+5V
-!R92
R117
PF3
R0K
1.00K
Rge
10K
U49A
100K
MR1 37
057B
2
D057A
1
D51 DL4148
>
4
1.00:
U50
D63
MAX90i
OL4148
R129
INH
PCI
CIA
PC2
CIS
Rb
2
13
514
10K
RFout3
R120
tO
U498
C145-E
560pF
C=
156
MM6D353
08
058BSA
7
PL13
1.00K
a
8
0.1pF
OD58A
Phase Clock3
-::- 10.0K
T
R70
C147
0.lpP
3
Vco in
Demout
R2
10
R99
24.32C
I
II
365K
15
K
1.
-=ot
R
.04
1.00K
14Sigin
A69
MAX901
]
R66
15
.
-
-10.0Kp
Vcoo
~Comp
Vo
J
RF CLOCK3
'
ShutdownX
1074AC08
74HCT904BA
R134
0
-
RF CLOCKr
4
RF CLOCK
2
1
Y
10.0K
iCI
1C2
1C3
ECO: Add Component
R4100
1.00K
2C0
2Y
2
132C2
2C3
2-
D52
DL4148
X0
U52
5INH
1
D64
L.20
74ACT253
ECO: LiM pis ISIS
and add jumpers
+5V
R121
100K
MMB0 35
pvF
T
713C1.00K
C
CI
APhase
Clock4
E
R73
U49C
-=::-10.0K
RIOS
Si-n
12
R2
CIAHCT946
3
MAX901
RI123
10K
MMBD 353
R124
212365K
RI
O.IpF
4C85
R 4
24.3
COM
C4
9
Demout10
_
CLOCK4
74ACOS
R15Fff
Brigham and Wo men's Hospital
Department of Radiology/MRI
LMRC. Room 007
13
ECO: Add Component
"
221 Longwood Ave.
Boston, MA 02115
0.0K
till.
MAX901
Size
a
)ale:
A
RF
11~
hudw4
74HCT9046A
15
0606
SD60A
10K
f
Vco-in
U49D
RFout4
R72
14
11
1.00K
1
7&7 10 K
12
059
~0591:
PL14
C148
10
PF4
R122
10K
0
"=~
RI02
10K
DL414_
2N
PCI
PFA14
Shutdown3
S ;hntdown4
4-
+5v
7
c
Phase Correction. Ch 3
Document
&4
Number
Ia. Doc)
,$aturday,
eC
Decombor
02, 199hoo
3
of
12
7
3
4
5
a
+7V
ECO:Add Components
O.-01pF
3.pF
U
R13
Shuldownl
-
0D5
100
RF CLOCK
1
D14148
7
2
F6345
0.IpF
I
014148I
RI
1.00
1
10.2pH
7.4pH
T2
TTSO-77
2:10
Ce
1s OOpF
d CB
02
RF634
j0.1p
L2
GRFout1
200pF
1C2A
L1
C4
FTSO-77
20pFs
O.lpF R2
1.00
74AC04
MC3415
200pF
DL4148
R3
so
Note.: Use
two
ohm
reshlrors
84
0
Note: Use to
.__
IO0ohm
resistors
a
+48VL
D3
L
U3
0.01p
VSW
4
0Y6i
22pF
C12 IC
O.Ol pF1
SK36DICT
FB
DL4148
L4
LT107447pH
VIN
CIIF4+C9
:I
CO
220pF
B
D4
DL4148
_
R5
DS
22pH
DL4148
K
_12.1
UAA
R0
-
I
R7
Shutdownl
LMC6484
.C13F
C14
0.15pF
BCX70
07
Cl
R8
08
12.1K
014148
Cl
R9
DL4148
4-
<=Vreill
3
100
C
U41
Cis
7
5
0.1pF
RIO
lOOK
U4C
S.OlpF
CJVorl
LMC6484
Rif
10
10.0K
LMC6484
CHI Power
senE
Brigham and Women's Hospital
1)
1)
Department of Radlology/MRI
LMRC, Room 007
221 Longwood Ave.
Boston, MA 02115
iola
Amplifier Stage. Channel I
ze
s
8
ale:
1
1
2
1
2
1
4
+
5
1
a
I
Document Number
C
(DoC)
Tuesday, November 07, 1995
7
hee
I
of
S
12
2
3
5
--
1
71
- -T
+7V
ECO:Add Components
3.3pF
0.01pF
R13/\
Shutdown2
U5
A
00L6
D4148
RF CLOCK:
C
7
2
10
<=RFout2
T4
.
T5
T30-77
C210 -
24 0p
C22
.
1800pF
C23
l05
-
p3
74AC04
C24
RF634S
1.00
--
FT50-77
.A
200pF
R13
MC3415
7.4pH
10:10
U2B
.-
10.2pH
200pF
D9
DL4148
1.00
4
L6
04
0.1F
3
15
DL4148
r
R14
R15
50
Note: Use
Itwo lohm
resistoers
50
Not: Use
two 100ohm
resistors
9
a
+48V
us
LT10
VIN
C27
.
VSW
DL4148 +
LS
DL4148
R16
4
D14
C25
220pF
0.0lpF
D12
L7
47pH
C28
SK36DICT
F8
0.01lFp
C26
220pF
22pH
.
12.1K
D13
L4148
0
2
F
-~T
U7A
<
3
121
29
C
Vro[12
LMC6484
-C30
0.l5pF
RIO
Shutdown2
RCX7O
C1.00M
_
D15
Dig
Rig
0L4148
R20
100
K
.12.1
Ml
DL4148
U78
C31
X
0.1 pF
R21K
U7C
~
--4
,
21OpF
Vfor2
LMC8484
R22
10
10.0K
LMC6484
CH2 Power So
Brigham and Women's Hospital
Department of Radiology/MRI
LMRC. Room 007
221 Longwood Ave.
Boston, MA 02115
Amplifier Stage, Channel
1
2
13
1
4
+
5
3ieI
a
Document Number
{Doc)
oC
)&Is!:
November
-----------L
Tuesday,
07,
2
Metl
1099
I
5
Ol
a
1
3
2
4
a
.7V
C33
33
.
C34
T
pF
.O
pF
A
U8
RF
CLOCKT
2-RFa34s
0. 1pF
1.00
U2C
-
4
P
L10
10.2pH
7.4pH
---
51
a-,
4
200pF
D17
014148
R23
5 >
L9
C
2400
08
0.IpF
74AC04
I
C40
200pF
R24
MC341D
R25
50
Noe: Use
r.te rohs
D67
Shutdown3
EOADdmL41n48a
100
U9
Nate Use
two lO0alim
-reslitrs
meintra
C41
220pF
FB
DL4149
L12
022
SK36DICT
1
C44 JL
0.01pF
a
0 20
47pH
VSW
VIN
+
R26
50
ECO:Add Components
+448V
C43
FT50-77
04148
1.0
R137
0.OtpF
C38[TA
-
3[IRFE34S
RFout3
TT50-77
T7 C37
10:10
p[
C39
-
TB
z:
I
S
DL-4148
R27
C42
220pF
1321
22pH
K
-12.1
DL4148
UIOA
-4
V
<=vfefl3
3
0
R28 IC45
LMC6484
-OK
C46
00p
O.15pF
R29
Shutdown3
B9X70
R30
023
1.00M
owr StjO
DL4148
024
CN1
R31
100
OL4148
S12.1K
UIOB
C47
T
O.IpF
R32
C48p
___
-- =
Vfor3
LMC6484
R33
UI0C
8
10
10.OK
LMC6484
CH3 Power
SoE:>
Brigham and Women's Hospital
Department of Radiology/MRI
LMRC. Room 007
221 Longwood Ave.
Boston, MA 02115
fIll.
Amplifier Stage. Channel 3
Size
1
1
2
2
3
3
I4
1
4
4
a||
%
)a
6
0Ocumont Number
Tuesday,
November
07,
1995
o
heet
a
12
1
2
3
1
4
1
5
1
L13
L14
10.2pH
7.4pH
a
+7V
C50
C49
3.3pFT
0.01pF
A
o
U11
>l
RF CLOCK
1
010
3
R34
01446
1.001
U20
C52
200pF
020
I.IpF
9
-------- 4
IRF634S
-1
T0
C53
2400p
C54
-
1800pF
5F34S
4
T50-77
1010
-
C56
0.1 pF
FT50-77
200pF
026
74AC04
MC3415
100
DL4148
R36
R37
50.
50
138
D8
Note: Us:
two to
Shutdown4
100
-.
DL4148
U12
0.01pF
D27
ECO:Add Components
+48V
5VIN
4
VSW
D28
DL4146
DL4148
47pH
~
L16
---C8Oji
FB
220PF
Note: Use
I"toohm
resistors
resistors
a
LT1074
JRFout4
Tit
0.01Tp
SK36DICT
T
T
C58
220pF
22pH
--
R38
D29
12.1 K
DL4148
z
U53
R39
Cot1p
-
LMC6484
.1p
-=0=-
C62
0.15pF
1,-I
Shutdown4
1CX70
1.OOM
c
-
D32
R41
031
L4148
R42
100
S12.1
K
c
DL4148
U539
7
C53
0.lpF
1004K
UO3C
01pF
~EJVfor4
LMC6404
R44
8
9
10
10.0K
LMC6484
CH4 Power
SeC
Brigham and Women's Hospital
Department of Radiology/MRI
LMRC, Room 007
221 Longwood Ave.
Boston. MA 02115
Amplifier Stage, Channel 4
he
B
)ae:
3
4
5
2
I Document Number
(Doe)
ruesday, November
C
07, 1995
ee
7
o
12
10
- -
110
+5V
V
+5
--
- - - -
--
- -
~+C77
7 33pF
-
- 111
1114
+*
-
1122
111
--------
1119-
111
-45V
,
U128
S.5V5
C82
F
81
3.3pF
eo
T .01 pF
.01
C83
3.3tt
C4
G.011F
Z
- V -
- - - -
S
-
-
-
U5
I~;~.+5iPF
B
3.3pe
CBS
O.flo p
+C7
3.v
,
U130
. +9-5V
.p
-
-
45
. T 0.01VF 33p F'
0CS
0
. i VFZI
33
0F
a
9
.p
9
0.1p
+
IO0
1133
U32
U131
,~
.j
- !45U4
I
.
O 1t
- - - - - -
-2
3I
.3F T .lp
. .!36
+v
-------
:
U29
1 ;59
+5V
,
:+5V
U123
5i/
T 33p
""""'"""
]OOp
--
+C7
C7
'+S
TOOIP
--
1121
+5V
118-------- -- --
-
- -2
-
2o+i5F
7
11 . .- - - -
12
11
+oy
C7
C3 C4 C5
C7
F 3F3: 001 FT .p O.O"'IL
0.0 1
1
-
1134
3.3ptF
F 33F 0.0 1pF'
U37
--
0.01 p
-
0. 1
sot
-
C12v
TOli
F.jOF
1138
+5vj
OOtzF,
V2
1141
1140
U139
-
+5v
A
20011SpF
2001SF
jaO.ipF
VOD VCC +SV
V.
F
0oi.
T.F,
-
Uj44
'I-'
+59 *OV59
211F.
VEE
t
1143
U42
-+59
20011SF,
_
U47
~
151
T C134
)O1F
2T.iF,
2I01o,
C135
TfO.OtpF,
-5v
2
Brigham and Women's Hospital
Department of RadiotogyMRt
i~~l)LMRC, Room 007
221 Longwood Ave.
Boston. MA 02115
Decoupling Capacitors
SIze
A
I
I
-1Document Number
iate
D
2 -1
5
heat
12
ol
s
)I
I7 "
1
3.3
057
FJ
01
2
C4
.
C13p'
9
F.
3.3li
CSCgI
2
3.3LF
:31:
-ioTl
-
CIO ,I:
-
06+SV
Us
C1
0011
11]
0.08 pfy
.2v
3.3121
C1
0
+5v
U4D
12 ,
UIOD
12
t3 >_L4
U~143B
12
1313
>J4
LMC6484
0
9
CLK
LMC6484
-
-0
U53D
U7D
74AC74
12
14
F
14
>
LMC6484
ftp>
+5V
LMCB484
E1HC32
-1
00
2
I-
-A
I
10. APPENDIX C
10.1 Pressure Calculations for the Phased Array (Zemanek 1971)
The pressure field from an ultrasound element through a lossy substrate can be
modeled as:
p(r)=
ip0 ck
2z
e-ik''~'
e-pd
,f
dA
r-r
where po is the tissue density, c is the speed of sound, k is the wavenumber (2r/X where X
is the ultrasonic wavelength), r is the coordinate vector (x,y,z) of the pressure poin, r is
the coordinate vector of the incremental source area of the complete transducer area A, A
is the attenuation coefficient in the lossy material (absorption and scattering), d is the ray
distance in the lossy material between the source point and the location of the desired
pressure point, and u is the complex surface velocity of that source (magnitude and
phase). For an array with N elements, the pressure at a given point m corresponding to a
location (xm, ym, zm) is given by:
Pm
ipck
N
n=1
eikrni e-pd UndA
rmn
where subscript n corresponds to the driving signal of the nth element of the array. This
model neglects temperature,
non-linear, refraction, and scattering effects on pressure
calculation (scattering attenuation lumped with absorption).
A.2 Pseudoinverse Technique (Ebbini and Cain 1989)
The pressure at a set of points can be derived from a transfer function hn, equal to
219
hm
=ipckf e-ikl e-" dA
z
2
A
rmn
such that
p = Hu
where p is an m x 1 vector corresponding to a set of pressures at m different spatial
locations, H is the m x n transfer matrix, and u is the n x 1 vector corresponding to the
driving velocities. This matrix can be inverted such that given a set of desired pressures
at given locations, the driving signals can be calculated. This can be accomplished using
the pseudoinverse. The matrix form of the transfer function can be written as:
H = XSY*
where X and Y are unitary matrices and S is a rectangular matrix with diagonal elements
corresponding to the eigenvalues of H. The pseudoinverse is then given by:
A
H =YSX*
where
+
indicates pseudoinverse, * indicates the complex conjugate transpose and S^ is
formed from S with the diagonal elements replaced by their reciprocal.
10.2 Intensityand Specific Absorption Rate Calculations (Hynynen 1990)
The time average power absorbed <q> by the tissue at location (x,y,z) is (when
the effects of shear viscosity are small for a CW, monofrequency signal) can be modeled
as:
< q(x,y,z) >= a
p2 (x,y,z)
POv
220
where a is the absorption coefficient, po is the tissue density, and v is the speed of sound
in the tissue.
10.3 Bioheat Transfer Equation (Pennes 1948)
The tissue temperature response can be simulated using the bioheat transfer
equation:
cdT(xy,z,t)
dt
=
kV 2T(x,y,z,t)
-
wcb(T(x,y,z,t)
-
T) +<q(x,y,z,t)>
where po is the tissue density, ct is the specific heat of the tissue, Cb is the specific heat of
the blood, k is the thermal conductivity, w is the perfusion, Ta is the arterial blood
temperature, and T(x,y,z,t) is the temperature at location (x,y,z) at time t. P2 is the spatial
gradient and d/dt is the time derivative.
10.4 Thermal Dose Calculation (Sapareto and Dewey 1984)
The thermal lesions are predicted using a numerical integration of the SaparetoDewey model:
D(t) =
R'(t)"'' dt,
where
R=2 if T(t)>430 C, 4 if T(t)<430 C.
221
11. REFERENCES
American Radio Relay League. 1989. Digital PEP Wattmeter and SWR Calculator. The
ARRL Handbook for the Radio Amateur:34-10-34-15.
Anderson EM, Electric Transmission Line Fundamentals, Prentice Hall, Reston, VA,
1995.
Ang SS, Power Switching Converters, Marcel Dekker, New York, 1995.
Avratoglou CP, Vougaris NC, Ioannidou Fl, "Analysis and Design of a Generalized
Class E Tuned Power Amplifier," IEEE Transactions on Circuits and Systems,
vol. 36, no.8, pp. 1068-1079,1989.
Barnard JW, Fry WJ, Fry FJ, Brennan JF, "Small localized ultrasonic lesions in the white
and gray matter of the cat brain.," Archives of Neurological Psychiatry, vol. 75,
pp. 15-35, 1956.
Basauri L, Lele PP, "A simple method for production of trackless focal lesions with
focused ultrasound: Statistical evaluation of the effects of irradiation on the
central nervous system of the cat," JPhysiol,vol. 160, pp. 513-534, 1962.
Baxandall PJ, "Transistor Sine-Wave LC Oscillators: Some General Considerations and
New Development," ProcInst Elec Eng, vol. 106, no.B, pp. 748-758, 1959.
Benkeser PJ, Frizzell LA, Ocheltree KB, Cain CA, "A tapered phased array ultrasound
transducer for hyperthermia treatment," IEEE Trans Ultrason FerroelectrFreq
Contr, vol. 34, pp. 446-453, 1987.
Bihrle R, Foster RS, Sanghvi NT, Fry FJ, Donohue JP, "High-intensity focused
ultrasound in the treatment of prostate tissue," Supplement to Urology, vol. 43,
no.2, pp. 21-26, 1994.
"Effects of physical parameters on high
Billard BE, Hynynen K, Roemer RB,
temperature ultrasound hyperthermia.," Ultrasound Med Biol, vol. 16, pp. 409420, 1990.
Blumgart LH, Fong Y, "Surgical options in the treatment of hepatic metastasis from
colorectal cancer," Curr Prob Surg, pp. 340-355+, 1995.
Born N, Lancee CT, Honkoop J, "Ultrasonics viewer for cross-sectional analysis of
moving cardiac structures," Biomedical Engineering,vol. 6, p. 500, 1971.
222
Botros YY, Ebbini ES, Volakis JL, "Two-step hybrid virtual array-ray (VAR) technique
for focusing through the rib cage," IEEE Trans Ultrason FerroelectrFreq Contr,
vol. 45, no.4, pp. 989-999, 1998.
Botros YY, Volakis JL, VanBaren P, Ebbini ES, "A hybrid computational model for
ultrasound phased-array heating in presence of strongly scattering obstacles,"
IEEE Trans Biomed Eng, vol. 44, no.11, pp. 1039-1050, 1997.
Bowman HF, "Heat transfer concepts and tumor heating," In Third International
Symposium on Cancer Therapy by Hyperthermia, Drugs, and Radiation, Journal
of the National CancerInstitute Fort Collins, CO,June 22,1980, pp.437-45, 1981.
Bowman HF, Thermodynamics of tissue heating:modeling and measurements for
temperature distributions, In Medical Aspects of Hyperthermia, ed. Nussbaum
GH, American Institute of Physics, 1982.
Bowman HF, Estimation of tissue bloodflow, In Heat Transfer in Medicine and Biology,
ed. Eberhart RC, Shitzer A, Plenum Press, New York City, 1984.
Bowman HF, Hyperthermia in Cancer Treatment, In Thermal Dosimetry, ed. Anghlieri
U, Robert J, CRC Press, Boca Raton, FL, 1986.
Bowman HF, Cravallo EG, Woods M, "Theory, measurement, and application of thermal
properties of biomaterials," Annual Review ofBiophysics and Bioengineering,vol.
4, pp. 43-80, 1975.
Bowman HF, Newman KD, Martin GT, Hsu M, Ibrahim S, Bodenhofer MF, Hansen JL,
Svensson GK, and Bornstein BA, "Routine quantification of blood perfusion
during clinical hyperthermia," In Proceedings of the 43rd Annual Meeting of the
Radiation Research Society and the 15th Annual Meeting of the North American
Hyperthermia Society San Jose, CA,Apr. 1,1995, 1995.
Bruix J, "Treatment of hepatocellular carcinoma," Hepatology, vol. 25, no.2, pp. 259262, 1997.
Buchanan MT, Hynynen K, "The design and evaluation of an intracavitary ultrasound
phased array for hyperthermia," IEEE Trans Biomed Eng, vol. 41, no.12, pp.
1178-1187, 1994.
Burov AK, "High intensity ultrasonic oscillations for the treatment of malignant tumors
in animal and man," Dokl Akad Nauk SSSR, vol. 106, pp. 239-241, 1956a.
Burov AK, Adreevskaya G, "The effect of ultra-acoustic oscillation of high intensity on
malignant tumors in animals and man," Dokl Akad Nauk SSSR, vol. 106, pp. 445448, 1956b.
223
Cain CA, Umemura SA, "Concentric-ring and sector vortex phased array applicators for
ultrasound hyperthermia therapy," IEEE Trans Microwave Theory Tech, vol.
MTT-34, pp. 542-551, 1986.
Carstensen EL, Miller M, Linke CA, "Biological effects of ultrasound," Jour Biol Phys,
vol. 2, pp. 173-192, 1974.
Challande P, "Optimizing ultrasonic transducers based on piezoelectric composites using
finite-element method," IEEE Trans Ultrason FerroelectrFreq Contr, vol. 37,
no.2, pp. 135-140, 1990.
Chapelon JY, Faure P, Plantier M, Cathignol D, Souchon R, Gorry F, Gelet A, "The
feasibility of tissue ablation using high intensity electronically focused
ultrasound," IEEE Ultrasonics Symp, vol. 93CH3301-9, pp. 1211-1214, 1993.
Chen L, ter Haar GR, Hill CR, "Influence of ablated tissue on the formation of highintensity focused ultrasound lesions," Ultrasound Med Biol, vol. 23, no.6, pp.
921-931, 1997.
Chung A, Hynynen K, Cline HE, Colucci V, Oshio K, Jolesz F, "Optimization of spoiled
gradient-echo phase imaging for in vivo localization of focused ultrasound beam,"
Magn Reson Med, vol. 36, no.5, pp. 745-752, 1996a.
Chung A, Hynynen K, Cline HE, Colucci V, Oshio K, Jolesz FA. 1996b. Optimization
of spoiled gradient-echo phase imaging for in vivo localization for focused
ultrasound beam. [Abstract]. Proc.SMR 4th Meeting, ISSN 1065-9889 3:1738
Cirocco WC, Schwartzman A, Golub RW,
"Abdominal wall recurrence after
laparoscopic colectomy for colon cancer," Surgery, vol. 117, no.6, pp. 719-720,
1994.
Clarke PR, Hill CR, "Synergism between ultrasound and x-rays in tumor therapy," Br J
Radiol, vol. 43, pp. 97-99, 1970.
Cline HE, Hynynen K, Schneider E, Hardy CJ, Maier SE, Watkins RD, Jolesz FA,
"Simultaneous magnetic resonance phase and magnitude temperature maps in
muscle.," Magn Reson Med, vol. 35, pp. 309-315, 1996.
Cline HE, Hynynen K, Watkins RD, Adams WJ, Schenck JF, Ettinger RH, Freund WR,
Vetro JP, Jolesz FA, "A focused ultrasound system for MRI guided ablation,"
Radiology, vol. 194, pp. 731-737, 1995.
Cline HE, Schenck JF, Watkins RD, Hynynen K, Jolesz FA, "Magnetic resonance guided
thermal surgery," Magn Reson Med, vol. 31, pp. 628-636, 1993.
224
Coleman DJ, Lizzi FL, Driller J, Rosado AL, Chang S, Iwamoto T, Rosenthal D,
"Therapeutic ultrasound in the treatment of glaucoma," Ophthalmology, vol. 92,
pp. 339-346, 1985.
Cook J, "Digital Clock Phase Shifter without a Phase Locked Loop," IEEE Transactions
on Circuits and Systems-I Fundamental Theory and Applications, vol. 40, no.4,
pp. 278-283, 1993.
Crum LA, Hynynen K, "Sound therapy,"Physics World, vol. 9, no.8, pp. 28-33, 1996.
Curley SA, Davidson BS, Fleming RY, Izzo F, Stephens LC, Tinkey P, Cromeens D,
"Laparoscopically guided bipolar radiofrequency ablation of areas of porcine
liver," Surg Endos, vol. 11, no.7, pp. 729-733, 1997.
Damianou CA, Hynynen K, "Focal spacing and near-field heating during pulse high
temperature ultrasound therapy," UltrasoundMed Biol, vol. 19, no.9, pp. 777-787,
1993.
Damianou C, Hynynen K, "Near-field heating during pulsed high temperature ultrasound
hyperthermia treatment," UltrasoundMed Biol, vol. 19, pp. 777-787, 1993.
Damianou C, Hynynen K, Fan X, "Evaluation of accuracy of a theoretical model for
predicting the necrosed tissue volume during focused ultrasound surgery," IEEE
Trans UltrasonFerroelectrFreq Contr, vol. 42, pp. 182-187, 1995.
Darkazanli A, Hynynen K, Unger E, Schenck JF, "On-line monitoring of ultrasound
surgery with MRI," JMagRes Imag, vol. 3, pp. 509-514, 1993.
Daum DR, Buchanan MT, Fjield T, Hynynen K, "Design and evaluation of a feedback
based phased array system for ultrasound surgery," IEEE Trans Ultrason
FerroelectrFreq Contr, vol. 45, no.2, pp. 431-438, 1998.
Daum DR and Hynynen K, "Optimization of thermal dose using switching mode patterns
of a spherically shaped square element phased array," In 1996 IEEE Ultrasonics
Symposium Proceedings,IEEE, New York, NY, vol. 2, Nov. 3,1996, pp.1309-12,
1996.
Daum DR, Hynynen K, "Thermal dose optimization via temporal switching in ultrasound
surgery," IEEE Trans Ultrason FerroelectrFreq Contr, vol. 45, no.1, pp. 208215, 1998.
Davies SC, Hill AL, Holmes RB, Halliwell M, Jackson PC, "Ultrasound quantitation of
respiratory organ motion in the upper abdomen," Br J Radiol, vol. 67, pp. 10961102, 1994.
225
Davis B and Lele PP, "A theoretical study of rapid hyperthermia by scanned focused
ultrasound," In The Winter Annual Meeting of the Americal Society of Mechanical
Engineers vol. 12, pp.51-8, 1989.
De Poorter J, "Noninvasive MRI thermometry with the proton resonance frequency
method: Study of susceptibility effects," Magn Reson Med, vol. 34, pp. 359-367,
1995.
De Silets CS,
Transducer arrays suitable for acoustic imaging, Dissertation Thesis,
Stanford University, 1978.
Delannoy J, Chen CN, Turner R, Lewin RL, Le Bihan D, "Noninvasive temperature
imaging using diffusion MRI," Magn Reson Med, vol. 19, pp. 333-339, 1991.
Dickinson RJ, Hall AS, Hind AJ, Young IR, "Measurement of changes in tissue
temperature using MR imaging," J Comp Ass Tomogr, vol. 10, pp. 468-472, 1986.
Do-Huu JP, Hartemann P, "Annular array transdcuer for deep acoustic hyperthermia,"
IEEE UltrasonicsSymp, vol. 81, no.CH1689-9, pp. 705-710, 1981.
Doci R, Gennari L, Bignami P, Montalto F, Morabito A, Bozzetti F, Bonalumi MG,
"Morbidity and mortality after hepatic resection of metastases from colorectal
cancer," Br JSurg, vol. 82, pp. 377-381, 1995.
Dorr LN, Hynynen K, "The effect of tissue heterogeneities and large blood vessels on the
thermal exposure induced by short high power ultrasound pulses," Int J
Hyperthermia, vol. 8, pp. 45-59, 1992.
Duck FA, Perkins MA,
"Amplitude-Dependent Losses in ultrasound exposure
measurement," IEEE Trans Ultrason FerroelectrFreq Contr, vol. 35, pp. 232241, 1988.
Eaton MD, Melen RD, Meindl JD, A flexible, real time system for experimentation in
phased arrayultrasoundimaging, In Acoustic Imaging, ed. Metherell AF, 1980.
Ebbini ES, Cain CA, "Multiple-focus ultrasound phased-array pattern synthesis: Optimal
driving-signal distributions for hyperthermia," IEEE Trans Ultrason Ferroelectr
Freq Contr, vol. 36, pp. 540-548, 1989.
Ebbini ES, Cain CA, "A spherical-section ultrasound phased array applicator for deep
localized hyperthermia," IEEE Trans Biomed Eng, vol. 38, pp. 634-643, 1991 a.
Ebbini ES, Cain CA, "Experimental Evaluation of a Prototype Cylindrical Section
Ultrasound Hyperthermia Phased-Array Applicator," IEEE Trans Ultrason
FerroelectrFreq Contr,vol. 38, no.5, pp. 510-520, 1991b.
226
Ebbini ES, Cain CA, "Optimization of the intensity gain of multiple-focus phased array
heating patterns," Int JHyperthermia,vol. 7, no.6, pp. 953-973, 1991c.
Ebbini ES, Umemura S-I, Ibbini M, Cain C, "A cylindrical- section ultrasound phased
array applicator for hyperthermia cancer therapy,"
IEEE Trans Ultrason
FerroelectrFreq Contr, vol. 35, pp. 561-572, 1988.
Fallone BG,Moran PR, Podgorsak EB, "Non-invasive thermometry with clinical X-ray
CT scanner," Med Phys, vol. 9, pp. 715-721, 1982.
Fan X,
Non-Invasive Ultrasound Surgery Using Spherically Curved Phased Arrays,
Dissertation Thesis, University of Arizona Dept. of Electrical and Computer
Engineering, 1995a.
Fan X, Hynynen K, "Control of the necrosed tissue volume during noninvasive
ultrasound surgery using a 16 element phased array," Med Phys, vol. 22, pp. 297308, 1995b.
Fan X, Hynynen K, "A study of various parameters of spherically curved phased arrays
for noninvasive ultrasound surgery," Phys Med Biol, vol. 41, pp. 591-608, 1996a.
Fan X, Hynynen K, "Ultrasound surgery using multiple sonications - treatment time
considerations," UltrasoundMed Biol, vol. 22, no.4, pp. 471-482, 1996b.
Fjield T, Fan X, Hynynen K, "A parametric study of the concentric-ring transducer
design for MRI guided ultrasound surgery," JAcoust Soc Am, vol. 100, pp. 12201230, 1996.
Fjield T, Hynynen K, "The Combined Concentric-Ring and Sector-Vortex Phased Array
for MRI Guided Ultrasound Surgery," IEEE Trans Ultrason FerroelectrFreq
Contr, vol. 44, no.5, pp. 1157-1167, 1997.
Fjield T, McDannold N, Silcox C, and Hynynen K, "In vivo verfication of the acoustic
model used to predict temperature elevations for MRI guided ultrasound surgery,"
In Proceedingsof the 1998 IEEE InternationalUltrasonicsSymposium 1998.
Fong Y, Blumgart LH, Cohen AM, "Surgical Treatment of Colorectal Metastases to the
Liver," CA: Cancer J Clin, vol. 45, pp. 50-62, 1995.
Frizzell LA, "Threshold dosages for damage to mammalian liver by high intensity
focussed ultrasound," IEEE Trans Ultrason FerroelectrFreq Contr, vol. 35, pp.
578-581, 1988.
Frizzell LA, Linke CA, Carstensen EL, Fridd CW, "Thresholds for focal ultrasonic
lesions in rabbit kidney, liver and testicle," IEEE Transactions on Biomedical
Engineering,vol. BME-24, pp. 393-396, 1977.
227
Frizzell LA, Benkeser PJ, Ocheltree KB, Cain CA, "Ultrasound phased arrays for
hyperthermia treatment.," IEEE UltrasonicsSymp, vol. 2, pp. 931-935, 1985.
Fry FJ, "Ultrasound visualization and modification of brain tissue," IEEE Trans Sonics
Ultras, vol. SU-17, no.3, 1970.
Fry FJ, "Intense focused ultrasound in medicine. Some practical guiding physical
principles from sound source to focal site in tissue," Eur Urol, vol. 23, no.suppl.
1, pp. 2-7, 1993.
Fry FJ, Brennan JF, Barnard JW, "Histological study of changes produced by ultrasound
in the gray and white matter of the central nervous system," UltrasoundMed Biol,
vol. 3, pp. 110-130, 1957.
Fry FJ, Johnson LK, "Tumor irradiation with intense ultrasound," UltrasoundMed Biol,
vol. 4,pp. 337-411, 1978.
Fry WJ, "Intracranial anatomy visualized in vivo by ultrasound," Invest Radiol, vol. 3,
no.4, 1968.
Fry WJ, Intracranial anatomy and ultrasonic lesions visualized by ultrasound, In
Ultrasonographia Medica, ed. Bock J, Ossoinig K,
Verlag der Wiener
Medizinischen Akademie, Vienna, 1971.
Fry WJ, Barnard JW, Fry FJ, Krumins RF, Brennan JF, "Ultrasonic lesions in the
mammalian central nervous system," Science, vol. 122, pp. 517-518, 1955.
Fry WJ, Wulff VJ, Tucker D, Fry FJ, "Physical factors involved in ultrasound induced
changes in living systems: I. Identification of non-temperature effects," JAcoust
Soc Am, vol. 22, pp. 867-871., 1950.
Fry WJ, Fry FJ, "Fundamental neurological research and human neurosurgery using
intense ultrasound.," IRE Trans Med Electron, vol. ME-7, pp. 166-181, 1960.
Fry WJ, Mosberg W, Barnard JW, Fry FJ, "Production of focal destructive lesions in the
central nervous system with ultrasound.," Journal of Neurosurgery, vol. 11, pp.
471-478, 1954.
Gelet A, Chapelon JY, Margonari J, Theillere Y, Gorry F, Souchon R, Bouvier R, "Highintensity focused ultrasound on human benign prostatic hypertrophy," Eur Urol,
vol. 23, pp. 44-47, 1993.
Goss SA, Frizzell LA, Kouzmanoff JT, Barich JM, Yang JM, "Sparse random ultrasound
phased array for focal surgery," IEEE Trans UltrasonFerroelectrFreq Contr, vol.
43, no.6, pp. 1111-1121, 1996.
228
Gough IR, "Quality of live as an outcome variable in oncology and surgery," Australian
& New ZealandJournalof Surgery, vol. 64, no.4, pp. 227-235, 1994.
Granberg HO, "Applying power MOSFETs in class D/E RF power amplifier design," RF
Design, 1985.
Grebenkemper J, "The tandem match--An accurate directional wattmeter," QST, pp. 1826, 1988.
Grebenkemper J, "Calibrating diode detectors," QST, pp. 3-8, 1990.
Gururaja TR, Newnham RE, Klicker KA, Schulze WA, Shrout TR, and Bowen LJ,
"Composite piezoelectric transducers," In 1980 Ultrasonics Symposium
Proceedings vol. 2, pp.576-81, 1980.
Gururaja TR, Schulze WA, Cross LE, and Newnham RE, "Ultrasonic properties of
piezoelectric PZT rod-polymer composites," In IEEE 1984 Ultrasonics
Symposium Proceedingspp.533-8, 1984.
Gururaja TR, Schulze WA, Cross LE, Newnham RE, "Piezoelectric composite materials
for ultrasonic transducer applications: evaluation of ultrasonic medical
applications," IEEE trans Sonics and Ultrasonics, vol. SU-32, no.4, pp. 499-513,
1985a.
Gururaja TR, Schulze WA, Cross LE, Newnham RE, Auld BA, "Piezoelectric composite
materials for ultrasonic transducer applications. I. Resonant modes of vibration of
PZT rod-polymer composites," IEEE trans Sonics and Ultrasonics, vol. SU-32,
no.4, pp. 481-498, 1985b.
Gururaja TR, Schulze WA, Cross LE, Newnham RE, Auld BA, Wang YJ, "Piezoelectric
composite materials for ultrasonic transducer applications. I. resonant modes of
vibration of PZT rod-polymer composites," IEEE Trans Sonics Ultras, vol. SU32, no.4, pp. 491-498, 1985c.
Hall LD, Talagala SL, "Mapping of PH and temperature distribution using chemicalshift- resolved tomography," JMagn Reson, vol. 65, pp. 501-505, 1985.
Heimburger RF, "Ultrasound augmentation of central nervous system tumor therapy,"
Indiana Medicine, vol. 78, pp. 469-476, 1985.
Holcomb GW3, Tomita SS, Haase GM, Dillon PW, Newman KD, "Minimally invasive
surgery in children with cancer," Cancer, vol. 76, no.1, pp. 121-128, 1995.
Horvath J, "Ultraschallwirkung beim menshlichen sarkom," Strahlentherapie,vol. 75, p.
119, 1944.
229
Houghton AW, Brennan PV, "Phased Array Control Using Phased Lock Loop Phase
Shifters," IEE Proceedings-H,vol. 139, no.1, pp. 3 1-37, 1992.
Hughes KS, Simon R, Songhorabodi S, et.al., "Resection of the liver for colorectal
carcinoma metastases: A multi-institutional study of indications for resection,"
Surgery, vol. 103, no.3, pp. 278-288, 1988.
Hunt JW, Lalonde R, Ginsberg H, Urchuk S, Worthington A, "Rapid heating: Critical
theoretical assessment of thermal gradients found in hyperthermia treatments," Int
JHyperthermia,vol. 7, pp. 703-718, 1991.
Hutchinson EB, Buchanan MT, Hynynen K, "Design and optimization of an aperiodic
ultrasound phased array for intracavitary prostate thermal therapies," Med Phys,
vol. 23, no.5, pp. 767-776, 1996a.
Hutchinson EB, Hynynen K, "Intracavitary phased arrays for non-invasive prostate
surgery," IEEE Trans Ultrason FerroelectrFreq Contr, vol. 43, no.6, pp. 10321042, 1996b.
Hynynen K, Biophysics and technology of ultrasound hyperthermia, In Methods of
External Hyperthermic Heating, ed. Gautherie M, Springer-Verlag, New York,
1990.
Hynynen K, "The threshold for thermally significant cavitation in dog's thigh muscle in
vivo," UltrasoundMed Biol, vol. 17, pp. 157-169, 1991.
Hynynen K, "Focused ultrasound surgery guided by MRI," Science&Medicine, vol. 3,
no.5, pp. 62-71, 1996a.
Hynynen K, Chung A, Colucci V, Jolesz FA, "Potential adverse effects of high intensity
focused ultrasound exposure," Ultrasound Med Biol, vol. 22, no.2, pp. 193-201,
1996b.
Hynynen K, Chung A, Fjield T, Buchanan MT, Daum DR, Colucci V, Lopath P, Jolesz F,
"Feasibility of using ultrasound phased arrays for MRI monitored noninvasive
surgery," IEEE Trans Ultrason FerroelectrFreq Contr, vol. 43, no.6, pp. 10431053, 1996c.
Hynynen K, Damianou CA, Colucci V, Unger E, Cline HE, Jolesz FA, "MR monitoring
of focused ultrasonic surgery of renal cortex: Experimental and simulation
studies," JMag Res Imag, vol. 5, pp. 259-266, 1995.
Hynynen K, Darkazanli A, Damianou C, Unger E, Schenck JF, "The usefulness of
contrast agent and GRASS imaging sequence for MRI guided noninvasive
ultrasound surgery," Invest Radiol, vol. 29, pp. 897-903, 1994.
230
Hynynen K, Darkazanli A, Unger E, Schenck JF, "MRI-guided noninvasive ultrasound
surgery," Med Phys, vol. 20, pp. 107-115, 1993.
Hynynen K, Freund WR, Cline HE, Chung AH, Watkins RD, Vetro JP, Jolesz FA, "A
clinical noninvasive MRI monitored ultrasound surgery method," RadioGraphics,
vol. 16, no.1, pp. 185-195, 1996.
Hynynen K, Jolesz FA. 1998. Demonstration of noninvasive ultrasound brain therapy
through intact skull. Ultrasound Med.Biol. [Forthcoming].
Hynynen K, Roemer R, Anhalt D, Johnson C, Xu ZX, Swindell W, Cetas TC, "A
scanned focussed multiple transducer ultrasonic system for localized hyperthermia
treatments," Int JHyperthermia,vol. 3, pp. 21-35, 1987.
Hynynen K, Shimm D, Anhalt D, Stea B, Sykes H, Cassady JR, Roemer RB,
"Temperature distributions during clinical scanned, focussed ultrasound
hyperthermia treatments," Int JHyperthermia,vol. 6, pp. 891-908, 1990.
Ibbini M, Ebbini ES, and Cain CA, "Ultrasound phased arrays for hyperthermia: new
techniques based on the field conjugation method," In Proc. IEEE 1987 Ultrason.
Symp. pp.863-6, 1987.
Jamison RL, Donohue JH, Nagorney DM, Rosen CB, Harmsen WS, Ilstrup DM,
"Hepatic resection for metastatic colorectal cancer results in cure for some
patients," Arch Surg, vol. 132, pp. 505-511, 1997.
Karrer HE, Dias JF, Larson JD, and Pering RD, "A phased array acoustic imaging system
for medical use," In 1980 Ultrasonics Symposium Proceedings vol. IEEE
80CH1602-2, pp.757-62, 1980.
Kazimierczuk M, "Collector Amplitude Modulation of the Class E Tuned Power
Amplifier," IEEE Transactions on Circuits and Systems, vol. CAS-31, no.6, pp.
543-549, 1984.
Kazimierczuk M, "Class E Tuned Power Amplifier with Nonsinusoidal Output Voltage,"
IEEE Transactionson Circuits and Systems, vol. SC-21, no.4, pp. 575-581, 1986.
Kazimierczuk M, "Phase-Modulated Series-Paralell Resonant Converter with Series
Load," IEE Proceedings,vol. 140, no.5, pp. 297-306, 1993.
Kazimierczuk M, Tabisz WA, "Class C-E High-Efficiency Tuned Power Amplifier,"
IEEE Transactionson Circuits and Systems, vol. 36, no.3, pp. 421-428, 1989.
Kinne DW, "Conservation surgery for breast cancer: selection criteria and technical
consideration," Surgery Today, vol. 24, no.9, pp. 767-771, 1994.
231
Kishi M, Mishima T, Itakura T, Tsuda K, and Oka M, "Experimental studies of effects of
intense ultrasound on implantable murine glioma," In Proceedings of the 2nd
European congress on ultrasonics in medicine, Exerpta Medica, Amsterdam,
pp.28-33, 1975.
Koisumi H, Suetsugu T, Fujii M, Shinoda K, Shinsaku M, Iked K, "Class DE HighEfficiency Tuned Power Amplifier," IEEE Transactionson Circuits and Systems,
vol. 43, no.1, pp. 51-60, 1996.
Kremkau FW, "Cancer therapy with ultrasound: a historical review," Journal Clinical
Ultrasound,vol. 7, pp. 287-300, 1979.
Kuroda K, Abe K, Tsutsumi S, Ishihara Y, Suzuki Y, Sato K, "Water proton magnetic
resonance spectroscopic imaging.," Biomedical Thermology, vol. 13, no.4, pp. 4362, 1995.
Lele PP, "A simple method for production of trackless focal lesions with focused
ultrasound: Physical factors," JPhysiol, vol. 160, pp. 494-512, 1962.
Lele PP, "Production of deep focal lesions by focused ultrasound - current status,"
Ultrasonics,vol. 5, pp. 105-122, 1967.
Lele PP, Pierce AD. 1973. The thermal hypothesis of the mechanism of ultrasonic focal
destruction in organised tissues.Interaction of ultrasound and biological tissues.
Washington, DC: Bureau of Radiological Health. FDA 73-8008 BRH/DBE.p. 121.
Linke CA, Carstensen EL, Frizzell LA, Elbadawi A, Fridd CW, "Localized tissue
destruction by high-intensity focused ultrasound," Arch Surg, vol. 107, pp. 887891, 1973.
Livraghi T, Lazzaroni S, Meloni F, Torzilli G, Vettori C, "Intralesional ethanol in the
treatment of unresectable liver cancer," World J Surg, vol. 19, no.6, pp. 801-806,
1995.
Lovejoy A, Pedrick P, Doran A, Delchar TA, Mills JA, Stamm A, "A Novel 8-bit
Ultrasound Phased-Array Controller for Hyperthermia Applications," Ultrasonics,
vol. 33, no.1, pp. 69-73, 1995.
Lynn JG, Zwemer RL, Chick AJ, Miller AE, "A new method for the generation and use
of focused ultrasound in experimental biology," J Gen Physiol, vol. 26, pp. 179193, 1942.
Maas-Moreno R, Damianou CA, "Noninvasive temperature estimation in tissue via
ultrasound echo-shifts. Part I. Analytical model.," J Acoust Soc Am, vol. 100,
no.4, pp. 2514-2521, 1996a.
232
Maas-Moreno R, Damianou CA, Sanghvi NT, "Noninvasive temperature estimation in
tissue via ultrasound echo-shifts. Part II. In vitro study.," J Acoust Soc Am, vol.
100, no.4, pp. 2522-2530, 1996b.
Madersbacher S, Kratzik C, Szabo N, Susani M, Vingers L, Marberger M, "Tissue
ablation in bening prostatic hyperplasia with high intensity focused ultrasound,"
Eur Urol, vol. 23, pp. 39-43, 1993.
Madersbacher S, Pedevilla M, Vingers L, Susani M, Marberger M, "Effect of highintensity focused ultrasound on human prostate cancer in vivo.," CancerRes, vol.
55, pp. 3346-3351, 1995.
Malcolm AL, ter Haar GR, "Ablation of tissue volumes using high intensity focused
ultrasound.," UltrasoundMed Biol, vol. 22, no.5, pp. 659-669, 1996.
Marcos-Alvarez A, Jenkins RL, Washburn WK, Lewis WD, Stuart KE, Gordon FD, Kane
RA, Clouse ME, "Multimodality Treatment of Hepatocellular Carcinoma in a
Hepatobiliary Specialty Center," Archives ofSurgery, vol. 131, pp. 292-298, 1996.
McCall JL, Booth MW, Morris DL, "Hepatic cryotherapy for metastatic liver tumors," Br
JHos Med, vol. 54, no.8, pp. 378-381, 1995.
McDannold N, Hynynen K, Wolf D, Wolf G, Jolesz F, "MRI Evaluation of Thermal
Ablation of Tumors with Focused Ultrasound," J Mag Res Imag, vol. 8, pp. 91100, 1998a.
McDannold N, Jolesz FA, Hynynen K, "The determination of the optimal delay between
sonications during focused ultrasound surgery using MRI to monitor thermal
build-up in vivo," Radiology, vol. Accepted for publication, 1998b.
McGough RJ, Ebbini ES, Cain CA, "Direct Computation of Ultrasound Phased-Array
Driving Signals from a Specified Temperature Distribution for Hyperthermia,"
IEEE Trans BiomedEng, vol. 39, no.8, pp. 825-835, 1992.
McGough RJ, Kessler ML, Ebbini ES, Cain CA, "Treatment planning for hyperthermia
with ultrasound phased arrays," IEEE Trans Ultrason FerroelectrFreq Contr,
vol. 43, no.6, pp. 1074-1084, 1996.
McGough RJ, Wang H, Ebbini ES, Cain CA, "Mode Scanning: Heating Pattern
Synthesis with Ultrasound Phased Arrays," Int J Hyperthermia, vol. 10, no.3, pp.
433-442, 1994.
Montori A, "Minimally invasive surgery," Endoscopy, vol. 30, no.2, pp. 244-252, 1998.
Moritz AR, Henriques FCJr, "Studies of thermal injury. II. The relative importance of
time and surface temperature in the causation of cutaneous burns.," Am J Path,
vol. 23, pp. 695-720, 1947.
233
Moros EG, Hynynen K, "A comparison of theoretical and experimental ultrasound field
distributions in canine muscle tissue in vivo," UltrasoundMed Biol, vol. 18, no.1,
pp. 81-95, 1992.
Mulligan TH, Lynch D, Mulvin JM, Greene JM, Smith JM, Fitzpatrick JM, "High
intensity focused ultrasound in the treatment of benign prostatic hyperplasia," Brit
J Urol, vol. 79, pp. 177-180, 1998.
Murray JM, Oleszek GM, "Design consideration in class D MOS power amplifiers,"
IEEE Trans on Ind Elect and Control Instr, vol. IECI-26, no.4, pp. 211-218, 1979.
Nakamura K, Baba S, Saito S, Tachibana M, Murai M, "High-intensity focused
ultrasound energy for benign prostatic hyperplasia: clinical response at 6 months
to treatment using Sonablate 200(tm)," JEndourology,vol. 11, no.3, pp. 197-201,
1997.
Ngo FC, Ultrasonic Phased-Array Driver System, Thesis Thesis, University of Illinois,
1988.
Ngo FC, Ebbini ES, and Cain CA, "An Experimental Analysis of a Sector-Vortex Phased
Array Prototype," In IEEE 1989 UltrasonicsSymposium Proceedings,IEEE, New
York, NY, vol. 2, Oct. 3,1989, pp.999-1002, 1989.
Nielson HJ, "Detrimental effects of perioperative blood transfusion," British Journal of
Surgery, vol. 82, no.5, pp. 582-587, 1995.
ONeil HT, "Theory of focusing radiators," JAcoust Soc Am, vol. 21, no.3, pp. 516-526,
1949.
Ocheltree KB, Benkeser PJ, Frizzell LA, Cain CA, "An Ultrasonic Phased Array
Applicator for Hyperthermia," IEEE Trans Sonics Ultras, vol. SU-31, pp. 526531, 1987.
Oka M, "Surgical application of high-intensity focused ultrasound," Clin All Round(Jpn),
vol. 13, p. 1514, 1960.
Parker DL, "Applications on NMR imaging in hyperthermia: an evaluation of the
potential for localized tissue heating and noninvasive temperature monitoring,"
IEEE Trans Biomed Eng, vol. 31, pp. 161-167, 1984.
Passlick B, Born C, Thetter 0, "[Cost comparison of minimal invasive surgery vs.
standard operation exemplified by primary pneumothorax]," Langenbecks Archiv
fur Chirugie,vol. 114, pp. 1290-1292, 1997.
Pennes HH, "Analysis of tissue and arterial blood temperatures in the resting human
forearm.," JournalofApplied Physiology, vol. 1, no.2, pp. 93-122, 1948.
234
Perez CA, Scott C, Enami B, Homback NB, Sneed PK, Asbell SO, Janjan NA,
"Evaluation of 45 degrees C hyperthermia and irradiation. II. A phase I clinical
trial in humans by the Radiation Therapy Onconlogy Group," Am J Clin Onc, vol.
16, no.6, pp. 477-481, 1993.
Perras H, "Broadband power-tracking VSWR bridge," Ham Radio, 979.
Prat F, Centarti M, Sibille A, Fadil A, Henry L, Chapelon JY, Cathignol D,
"Extracorporeal high-intensity focused ultrasound for VX2 liver tumors in the
rabbit," Hepatology, vol. 21, no.3, pp. 832-836, 1995.
Prat F, Chapelon JY, Fadil A, Sibille A, Theilliere Y, Ponchon T, Cathignol D, "Focused
liver ablation by cavitation in the rabbit: a potential new method of extracorporeal
treatment," Gut, vol. 35, no.3, pp. 395-400, 1994.
Pressman AL, Switching Power Supply Design , McGraw-Hill, New York, 1998.
Raab FH, "Radio Frequency Pulsewidth Modulation," IEEE Transactionson Comm, vol.
53, pp. 958-966, 1973.
Raab FH, "Idealized Operation of the Class E Tuned Power Amplifier," IEEE
Transactionson Circuits and Systems, vol. CAS-24, no.12, pp. 725-735, 1977.
Ramshaw BJ, "Laproscopic surgery for cancer patients," CA: Cancer J Clin, vol. 47,
no.6, pp. 327-350, 1997.
Reid JM and Wild JJ, "Current developments in ultrasonic equipment for mechanical
diagnosis," In Proc.Nat. Electronics Councilvol. 12, pp.44-58, 1956.
Sanghvi NT, "Role of cavitation during high intensity focused ultrasound treatment of
prostate tissue," In Proceedings16th InternationalCongress on Acoustics and the
135th Meeting of the Acoustic Society ofAmericaJune 20,1998, pp.l1067-8, 1998.
Sanghvi NT, Fry FJ, Bihrle R, Foster RS, Phillips MH, Syrus J, Zaitsev AV, Hennige
CW,
"Noninvasive surgery of prostate tissue by high-intensity focused
ultrasound.," IEEE Trans Ultrason Ferroelectr Freq Contr, vol. 43, no.6, pp.
1099-1110, 1996.
Sanghvi NT, Hawes RH,
"High-intensity focused ultrasound," Experim Investig
Endoscopy, vol. 4, no.2, pp. 383-395, 1994.
Sanghvi NT, Hynynen K, Lizzi FL, "New developmnets in therapeutic ultrasound," IEEE
Trans Biomed Eng, vol. (submitted), 1996.
Sapareto SA, Dewey WC, "Thermal dose determination in cancer therapy," Int J
Radiation Oncology Biol Phys, vol. 10, pp. 787-800, 1984.
235
Sato M, Watanabe Y, Udeda S, Iseki S, Abe Y, Sato N, Kimura S, Okubo K, Onji M,
"Microwave coagulation therapy for hepatocellular carcinoma," Gastroenterology,
vol. 110, no.5, pp. 1507-1514, 1996 .
Scheele J, Stang R, Altendorf-Hofmann A, Paul M, "Resection of colorectal liver
metastases," World JSurg,vol. 19, pp. 59-71, 1995.
Seip R, VanBaren P, Cain CA, Ebbini ES,
"Noninvasive real-time multipoint
temperature control for ultrasound phased array treatments," IEEE Trans
UltrasonFerroelectrFreq Contr, vol. 43, pp. 1063-1073, 1996.
Senak PJr, "Amplitude Modulation of the Switched-Mode Tuned Power Amplifier,"
Proc IEEE, vol. 53, 1965.
Shrout TR, Bowen LJ, Schulze WA,
"Extruded PZT/polymer composites for
electromechanical transducer applications," Materials Research Bulletin, vol. 15,
no.10, pp. 1371-1379, 1980.
Sibille A, Prat F, Chapelon JY, Fadil A, Henry L, Theillere Y, Ponchon T, Cathignol D,
"Characterization of extracorporeal ablation of normal and tumor-bearing liver
tissue by high intensity focused ultrasound," UltrasoundMed Biol, vol. 19, no.9,
pp. 803-813, 1993.
Simon C, VanBaren P, Ebbini ES, "Two-dimensional temperature estimation using
diagnostic ultrasound," IEEE Trans Ultrason FerroelectrFreq Contr, vol. 45,
no.4, pp. 1088-1099, 1998.
Smith NB, Webb AG, Ellis DS, Wilmes LJ, OBrien WD, "Experimnetal verification of
theoretical in vivo ultrasound heating using cobalt detected magnetic resonance,"
IEEE Trans UltrasonFerroelectrFreq Contr, vol. 42, no.4, pp. 489-491, 1995.
Sokal NO, "Class E High-Efficiency Switching-Mode Tuned Power Amplifier with Only
One Inductor and One Capacitor in Load Network-- Approximate Analysis,"
IEEE Journalof Solid-State Circuits,vol. SC-16, no.4, pp. 380-384, 1981.
Sokal NO, Sokal AD, "Class E -- A New Class of High-Efficiency Tuned Single-Ended
Switching Power Amplifiers," IEEE Journal of Solid-State Circuits, vol. SC- 10,
no.3, pp. 168-176, 1975.
Somer JC, "Electronic sector scanning for ultrasonic diagnosis," Ultrasonics,vol. 6, pp.
153-159, 1968.
Sowlati T, Greshishchev Y, Salama CAT, Rabjohn G, and Sitch J, "Linear Transmitter
Design using High Efficiency Class E Power Amplifier," In Proceedings of 6th
International Symposium on Personal, Indoor and Mobile Radio
Communications, IEEE, New York, NY, vol. 3, Sept. 27,1995, pp.1233-7, 1995.
236
Sowlati T, Greshishchev Y, Salama CAT, Rabjohn G, and Sitch J, "Linearized High
Efficiency Class E Power Amplifier for Wireless Communications," In
Proceedings of the IEEE 1996 Custom Integrated Circuits Conference, IEEE,
New York, NYMay 5,1996, pp.201-4, 1996.
Spaulding W, "A broadband two-port s-parameter test set," Hewlett-PackardJournal,
1984.
Stepanow B, Huber P, Brix G, Debus J, Bader R, van Kaick G, Lorenz WJ, "Fast MRI
temperature monitoring: Application in focused ultrasound therapy of malignant
tissue in vivo.," Proc SMR 3rd Meeting, ISSN 1065-9889, p. 1172, 1995.
Stewart HF, Ultrasonic Measurement Techniques and Equipment Output Levels, In
Essentials of Medical Ultrasound: A Practical Introduction to the Principles,
Techniques, and Biomedical Applications, ed. Repacholi MH, Benwell DA,
Humana Press, Clifton, NJ, 1982.
Susani M, Madersbacher S, Kratzik C, Vingers L, Marberger M, "Morphology of tissue
destruction induced by focused ultrasound," Eur Urol, vol. 23, pp. 34-38, 1993.
Suzuki T, Fujimoto K, Aida S, Isihara Y, Watanabe H, Okamoto K, Iorita N, Shirai S,
Orikasa S, "MRI monitoring during high-intensity focused ultrasound treatment,"
Proc SMR 3rd Meeting, ISSN1065-9889, vol. 2,p.1177, 1995.
Szent-Gorgyi A, "Chemical and biological effects of ultrasonic radiation," Nature, vol.
131,p.278, 1933.
Tancrell RH, Callerame J, and Wilson DT, "Near-field transient acoustic beamforming
with arrays," In 1978 Ultrasonics Symposium Proceedingsvol. IEEE 78CH1344ISU, pp.339-43, 1978.
Taylor M, Forster J, Langer B, Taylor BR, Grieg PD, Mahut C, "A study of prognostic
factors for hepatic resection for colorectal metastases," Am J Surg, vol. 173, pp.
467-471, 1997.
ter Haar GR, "Ultrasound focal beam surgery.," UltrasoundMed Biol, vol. 21, no.9, pp.
1089-1100, 1995.
ter Haar GR, Clarke RL, Vaughan MG, Hill CR, "Trackless surgery using focused
ultrasound: technique and case report," Minimally Invasive Therapy, vol. 1, pp.
13-19, 1998a.
ter Haar GR, Rivens I, Moskovic E, Huddart R, and Visioli AG, "Phase 1 clinical trials
of the use of focused ultrasound surgery for the treatment of soft tissue tumours,"
In SPIE Photonics West 1998 1998b.
237
ter Haar GR, Sinnett D, Rivens I, "High intensity focused ultrasound - a surgical
technique for the treatment of discrete liver tumors," Phys Med Biol, vol. 34, pp.
1743-1750, 1989.
Thomas J-L, Fink MA, "Ultrasonic beam focusing through tissue inhomogeneities with a
time reversal mirror: application to transskull therapy.," IEEE Trans Ultrason
FerroelectrFreq Contr, vol. 43, no.6, pp. 1122-1129, 1996.
Thomenius KE, "Evolution of Ultrasound Beamformers," 1996 IEEE Ultrasonics
Symposium, pp.1615-22, 1996.
Thurstone FL, von Ramm OT, Electronic beam steering for ultrasonic imaging, In
Ultrasound in Medicine, ed. deVlieger M, American Elsevier Publishing, New
York, 1974.
Umemura S, Cain CA, "The sector-vortex phased array: acoustic fields synthesis for
hyperthermia," IEEE Trans Ultrason FerroelectrFreq Contr, vol. 36, pp. 249257, 1989.
Vallancien G, Chartier-Kastler E, Bataille N, Chopin D, Harouni M, Bougaran J,
"Focused extracorporeal pyrotherapy,"Eur Urol, vol. 23, pp. 48-52, 1993.
Vallancien G, Harouni M, Guillonneau B, Veillon B, Bougaran J, "Ablation of
superficial bladder tumors with focused extracorporeal pyrotherapy," Urology,
vol. 47, no.2, pp. 204-207, 1996.
Vallancien G, Harouni M, Veillon B, Mombet A, Prapotnich D, Bisset JM, Bougaran J,
"Focused extracorporeal pyrotherapy: Feasibility study in man," J Endourology,
vol. 6, pp. 173-180, 1992.
Van Natta TL, Morris JA, Eddy VA, Nunn CR, Rutherford EJ, Neuzil D, Jenkins JM,
Bass JG, "Effective bedside surgery in critically injured patients is safe and costeffective," Annals ofSurgery, vol. 227, no.5, pp. 618-624, 1998.
VanBaren P, Seip R, Ebbini ES,
"Real-time dynamic focusing through tissue
inhomogeneities during hyperthermia treatments with phased arrays," IEEE
UltrasonicsSymp, 94CH3468-6,ISSN: 1051-0117, pp. 1815-1819, 1994.
Vaughan MG, ter Haar GR, Hill CR, Clarke RL, Hopewell JW, "Minimally invasive
cancer surgery using focused ultrasound: a pre-clinical, normal tissue study," Br J
Radiol, vol. 67, pp. 267-274, 1994.
von Ramm OT, Thurstone FL, Thaumascan: improved image quality and clinical
usefulness, In Ultrasound in Medicine, Plenum Press, NY, NY, 1970.
238
Wan H, VanBaren P, Ebbini ES, Cain CA, "Ultrasound Surgery: Comparison of
strategies using phased array systems.," IEEE Trans Ultrason FerroelectrFreq
Contr, vol. 43, no.6, pp. 1085-1098, 1996.
Wang H, Ebbini E, Cain CA, "Effect of phase errors on field patterns generated by an
ultrasound phased-array hyperthermia applicator,"
IEEE Trans Ultrason
FerroelectrFreqContr, vol. 38, no.5, pp. 521-531, 1991.
Wells PNT, Biomedical Ultrasonics, Academic Press, Boston, 1977.
Woeber K, The effect of ultrasound in the treatment of cancer, In Ultrasonic energy:
biological investigations and medical applications, ed. Kelly E, University of
Illinois Press, Urbana, 1965.
Wu F. (personal communication), 1998.
Yamanaka N, Okamoto E, Tanaka T, Oriyama T, Fujimoto J, Furukawa K, Kawamura E,
"Laproscopic microwave coagulonecrotic therapy for hepatocellular carcinoma,"
Surg Lapr Endo, vol. 5, no.6, pp. 444-449, 1995.
Yang R, Reilly CR, Rescorla FJ, Faught PR, Sanghvi NT, Fry FJ, Franklin TD, Lumeng
L, Grosfeld JL,
"High-intensity focused ultrasound in the treatment of
experimental liver cancer.," Arch Surg, vol. 126, pp. 1002-1010, 1991.
Yang R, Sanghvi NT, Rescorla FJ, Kopecky KK, Grosfeld JL, "Liver cancer ablation
with extracorporeal high-intensity focused ultrasound," Eur Urol, vol. 23 (Suppl.
1), pp. 15-22, 1993.
Zahradnik RL, Theory and techniques of optimization for practicing engineers, Barnes &
Noble, New York, 1971.
Zemanek J, "Beam behavior within the nearfield of a vibrating piston," JAcoust Soc Am,
vol. 49, pp. 181-191, 1971.
239