Digital Mammography Update:
Design and Characteristics of
Current Systems
Outline
Digital imaging in MAP
Detector technologies
Manufacturers and their wares
Doug Pfeiffer, MS, DABR
Boulder Community Hospital
Self-Assessment Module (SAM)
To assist your efforts toward ABR MOC
Quiz at the end of the talk
SAM
–
–
–
–
–
–
CAMPEP (Cat 1) activity
20 / 10 years (1/year)
Instructional content
5 mult. choice questions
feedback
no passing requirement
Thanks to…
Tony Seibert, PhD
Eric Berns, PhD
Penny Butler, MS
John Sandrik, PhD (GE)
Magnus Olofsson (Sectra)
A real SAM!
1
FFDM and the ACR
MAP statistics show a growing adoption of
digital mammography
US Mammography Facilities
and Units
14000
13000
12000
# 11000
10000
9000
8000
7000
2000
2001
2002
2003 2004
Ye a r
# Facilities
As of July 1, 2007
2007
# Units
Facilities
22% of all mammography units
1991 facilities had FFDM units
–
2006
Full Field Digital Mammography in
the US
2926 FFDM units
–
2005
Units
23% of all US facilities
FFDM growing about 6% / month
2
Screen-Film vs. FFDM Trends in the
US
FFDM Manufacturers and Models
Undergoing ACR Accreditation
14000
80, 6%
12000
# Units
10000
GE SENOGRAPHE 2000 D
8000
GE SENOGRAPHE DS
564, 42%
412, 30%
6000
FISCHER SENOSCAN
4000
LORAD SELENIA
2000
SIEMENS MAMMOMAT
NOVATION DR
O
ct
Ja 03
n0
A 4
pr
-0
Ju 4
l-0
O 4
ct
Ja 04
nA 05
pr
-0
Ju 5
l-0
O 5
ct
Ja 05
nA 06
pr
-0
Ju 6
l-0
O 6
ct
Ja 06
n0
A 7
pr
-0
Ju 7
l-0
7
0
Screen-Film Units
222, 16%
Screen-Film and FFDM
Accreditation Results
#
Units*
ScreenFilm
FFDM
(1308 units completing accreditation
2/15/03 – 7/12/06)
Reasons Facilities Do Not Pass
Accreditation
Overall
Pass
14,574 88.7%
1711
82, 6%
FFDM Units
92.7%
Deficient
1st Attempt Screen-Film
Deficiencies
(2/15/03 - 7/12/06)
72.3%
1st Attempt FFDM
Deficiencies
(2/15/03 - 7/12/06)
74.0%
11.3%
7.3%
26.0%
27.7%
Phantom
Clinical
Phantom
Clinical
*1st attempt for both initials and renewals; 2/15/03 – 7/12/06
3
Clinical Images:
Fatty vs. Dense Deficiencies
Phantom Images and Dose
Average Scores
Ave Dose*
# Units Fibers Specks Masses (mrads)
FFDM Clinical
Deficiencies
Screen-Film Clinical
Deficiencies
52.4%
43.1%
ScreenFilm
(SD)
14,574
4.70
(0.48)
3.60
(0.4)
3.74
(0.41)
168.7
(31.4)
1711
4.84
(0.54)
3.85
(0.33)
4.00
(0.39)
128.6
(38.6)
56.9%
47.6%
Dense
Fatty
Dense
Fatty
FFDM
(SD)
*as measured by TLD
Detailed Information
Source
Digital Detector Technologies
Indirect
–
–
Cesium iodide with CCD
Cesium iodide with TFT
Direct
–
Amorphous selenium with TFT
Computed Radiography (CR)
Photon Counting
–
Crystalline silicon
1966?
4
Indirect – CsI with CCD
Theory: CsI with CCD (slot scan)
[Fischer SenoScan]
(Philips MammoDiagnost FD Eleva = Fischer?)
X-rays
CsI Scintillator
1-toto-1 Fiber
Optic Coupler
CCD Array
[ ] = no longer available, ( ) = not FDA approved
Implementation: CsI with CCD (slot
scan)
Cesium Iodide
Scintillator
Fiber Optic Coupling of CsI Scintillator to
Rectangular CCD Arrays
Fischer SenoScan System
Fiber Optic
Plate
4 CCD
Arrays
1 cm
wide slot
5
Fischer SenoScan Details
Fischer SenoScan Details
Detector size 21 x 29 cm (N)
(scanner area) 10.5 x 14.5
(HR)
X-Ray Tube
W target
Al filter
Grid
None
Pixel size
54 µm (N)
27 µm (HR)
Image size
4096 x 5625
pixels (34.6 MB)
DQE (Mo/Mo, 28 1 mm-1 – 24%
kVp, 8.5 mR=85
µGy)
3 mm -1 – 22%
5 mm -1 – 19%
Dynamic
Range
12 bits
Fischer SenoScan Details
fNy
10 mm-1
M(fNy)
0.13
Display time
12-15 seconds
Chest wall
<8.5 mm
missing tissue
Magnification
Digital
magnification
only
Fischer SenoScan System
Acquisition
delay
Automatic
Exposure
Control
Automatic,
prescan
6
Indirect – Csi with TFT
Theory: CsI with TFT
X-ray photons
GE Senographe Essential
GE Senographe DS
[GE Senographe 2000D]
Cesium Iodide (CsI)
Light Photons
Amorphous Silicon
Photodiode/Transistor Matrix
Electrons
Readout Electronics
1980
Digital Data
X-ray Photon
Theory: CsI with TFT
200-300 µm thick
90% absorption
CsI-Needles, 5 - 10µm
5 µm
Columns of
cesium iodide
scintillator
X-rays
Amorphous
silicon
detectors
Substrate
Pixel Size (i.e. 100 µm)
Amorphous Silicon Detector
7
Implementation: Silicon Diode
Array Detector
Contact Fingers
on 3 Edges
Contact Leads
For ReadRead-Out
Electronics
Amorphous
Silicon Array
Glass Substrate
Scintillator (CsI)
CsI)
GE Senographe Details
DS
Essential
19.2 x 23.0
19.2 x 23.0
24.0 x 30.7
Pixel size
100 µm
100 µm
100 µm
Image size
1914 x 2294
pixels (9 MB)
1914 x 2294
pixels (9 MB)
2394 x 3062
pixels (14 MB)
DQE (Mo/Mo, 28 1 mm-1 – 52%
kVp, 8.5 mR=85
3 mm -1 – 49%
µGy) (GE doc.)
-1
0 mm-1 – 50%
2 mm -1 – 41%
5 mm -1 – 15%
0 mm-1 – 60%
2 mm -1 – 52%
5 mm -1 – 18%
Dynamic
Range
14 bits
14 bits
DS
– 30%
GE Senographe Details
Essential
X-Ray Tube
Mo/Rh target
Mo/Rh filter
Mo/Rh target
Mo/Rh filter
Mo/Rh target
Mo/Rh filter
Grid
5:1
5:1, 31 lp/cm
5:1, 36 lp/cm
Chest wall
<4 mm
missing tissue
<4 mm
<5 mm
Magnification
1.5X, 1.8X
1.5X, 1.8X
1.5X, 1.8X
2000D
Detector size
5 mm
100 micron
pixel size
2000D
GE Senographe Details
2000D
mm-1
DS
Essential
5 mm-1
fNy
5
M(fNy)
0.37
(0.37)
(0.37)
Display time
10 seconds
<10 sec raw,
<15 sec
processed
16 sec raw
21 sec
processed
<18 seconds
<25 seconds
Acquisition
delay
Automatic
Exposure
Control
5
mm-1
AOP (CNT,
AOP (CNT,
STD, DOSE),
STD, DOSE),
auto dense loc. auto dense loc.
AOP (CNT,
STD, DOSE),
auto dense loc.
8
GE Senographe Systems
Amorphous Selenium with TFT
Hologic Selenia
Siemens Mammomat NovationDR
(Agfa DM 1000 – NovationDR)
(IMS Giotto Image MD/SD-SDL)
(Planmed Sophie Nuance)
Essential
DS
(2000D no
longer
available)
Theory: Amorphous Selenium with
TFT
Implementation: aSe Detector
9
Implementation: aSe Detector
aSe Unit Details
Detector
size
aSe Unit Details
Siemens
Novation
X-Ray
Tube
Mo target
Mo/Rh filter
Mo target
Mo/W target
Mo/Rh/Al filter Mo/Rh filter
Grid
Cellular
5:1, 31 lp/cm
≤5 mm
Magnifica- 1.8X
tion
1.5X, 1.8X
Siemens
Novation
Giotto Image Planmed
Nuance
24.0 x 29.0
24.0 x 29.0
24.0 x 30.0
24.0 x 30.0
(?)
Pixel size 70 µm
70 µm
85 µm
85 µm
Image
size
3328 x 4096
pixels (24 MB)
3328 x 4096
pixels (24 MB)
2816 x 3584
pixels (16 MB)
2816 x 3584
pixels (20 MB)
DQE
1 mm-1 – 56%
(40%)
(Mo/Mo, 28
kVp, 8.5
mR)
3 mm-1 – 44%
(30%)
Dynamic
Range
14 bits
5
mm-1
– 30%
1 mm-1 – 63%
3 mm -1 – 52%
5 mm -1 – 31%
(20%)
14 bits
13 bits
16 bits
aSe Unit Details
Hologic
Selenia
Chest wall
missing
tissue
Hologic
Selenia
Giotto Image Planmed
Nuance
Mo target
Mo/Rh filter
5:1, 36 lp/cm 5:1, 34 lp/cm
Hologic
Selenia
Siemens
Novation
Giotto Image Planmed
Nuance
7.14 mm-1
5.88 mm-1
fNy
7.14 mm-1
M(fNy)
0.44
Display
time
15-20
seconds
40-50
seconds
4 seconds
5-10
seconds
Manual cell
selection,
auto breast
density
detection
Manual cell
selection, auto
breast density
detection
Manual cell
selection
Auto cell
selection, auto
kVp selection
0.46
Acquisition
delay
1.8X
1.6X, 1.8X,
2.0X
Automatic
Exposure
Control
10
aSe Units
Cassette-based (CR) Systems
Fuji ClearView
(Philips PCR Eleva CosimaX)
(Kodak DirectView CR)
(Agfa CR MM3.0 Mammo)
Planmed Nuance
Siemens Novation
Theory: Computed Radiography
PSP
x-ray exposure
1960
Giotto Image
Theory: Computed Radiography
laser beam scan
plate readout:
extract latent image
Base support
light erasure
plate exposure:
create latent image
plate erasure:
remove residual signal
11
Theory: Computed Radiography
Standard resolution: ~100 µm BaFBr
High resolution: ~50-70 µm BaFBr
Dual-side read; structured phosphor, CsBr
8 eV
Implemetation: Computed
Radiography
f-theta
lens
Reference detector
Cylindrical m irror
Beam splitter
Light channeling guide
Laser
Source
Output Signal
Beam deflector
PSL
3 eV
F center trap
Laser
2 eV
stimulation
Incident xx-rays
Implemetation : Computed
Radiography
ADC
PMT
Laser beam:
Scan direction
Amplifier
To image
processor
Plate translation:
Sub-scan direction
Implemetation : Double-Sided
Readout
Incident Laser Beam
photodetector
Protective
layer
laser beam
Light guide assembly
and PMT
optical
guide
mirror
Phosphor
layer
Protective Layer
Light
Scattering
Photostimulated
Luminescence
Phosphor Layer
emission
Transparent
support
Laser Light Spread
Base Support
"Effective" readout diameter
optical
guide
photodetector
12
CR System Details
Photon Counting
Fuji FCRm
Philips PCR
Eleva
Kodak
DirectView
Agfa CR
MM3.0
18 x 24
24 x 30
18 x 24
24 x 30
18 x 24
24 x 30
18 x 24
24 x 30
Pixel size
50 µm
50-200 µm
50 µm
50 µm
fNy
10 mm-1
10 mm-1
10 mm-1
10 mm-1
M(fNy)
0.37
Image
size
3328 x 4096
pixels (24 MB)
4728 x 5928
pixels (xx MB)
2816 x 3584
pixels (MB)
4760 x 5840
pixels (42 MB)
Detector
size
Sectra MicroDose
DQE
Dynamic
Range
14 bits
Photon Counting Theory
12 bits
1953
Photon Counting Theory
13
Photon Counting Theory
Photon Counting Theory
Quoted 90% detection efficiency (edge-on detectors)
Quoted 94% scatter rejection
2 MHz/pixel maximum count rate
3.6 eV to create electron-hole pair
–
–
both Compton and PE contribute
~103 hp per photon
Custom-designed ASIC (Application Specific Integrated Circuit)
contains pre-amplifier, shaper, comparator, digital counter
–
–
–
–
100 nS pulse
2 MHz count rate
~200 electron rms noise
no Swank noise
Operates close to quantum limit
Sectra MicroDose Details
Sectra MicroDose Details
Detector size
(scanned area)
24 x 26 cm
X-Ray Tube
W target
Al filter
Pixel size
50 µm
Grid
None
Image size
4800 x 5200
pixels (37 MB)
Chest wall
missing tissue
<5 mm
DQE (Mo/Mo, 28
1 mm-1 – 60%
3 mm -1 – 40%
5 mm -1 – 18%
Magnification
Digital
magnification
only
kVp, 8.5 mR=85
µGy)
Dynamic
Range
12 bits
Monnin, et al
14
Sectra MicroDose Details
Display time
Sectra MicroDose System
<20 seconds
Acquisition
delay
Automatic
Exposure
Control
Automatic,
prescan
MTF Comparison
DQE Comparison
Monnin, et al
15
DQE Comparison
Dose Comparison (AGD)
Lazzari, et. al.
131 µGy
GE DS
(1.0 mGy)
GE 2000D
1.5 mGy
Fuji
1.9 mGy
Fischer
SenoScan
1.3 mGy
(1.5 mGy)
1.8 mGy
(2.1 mGy)
[0.5 mGy]
Hologic
Selenia
Sectra
MicroDose
Future Developments
Tomosynthesis (Lorad, GE, Planmed)
Dual energy
Contrast enhancement
CAD integration
??
Doses from:
Bloomquist et al
or
(measured clinically)
or
[manufacturer]
SAM questions
Follow on the next slides
Use response system
16
Of the following systems, which
incorporate CsI in the detector?
3%
0%
89%
3%
6%
1.
2.
3.
4.
5.
Of the following systems, which
incorporate CsI in the detector?
Lorad Selenia
Sectra Microdose
GE Essential
Siemens Mammomat Novation
Fuji FCRm
1.
2.
3.
4.
5.
Of the following systems, which
incorporate CsI in the detector?
1.
2.
3.
4.
5.
Lorad Selenia – amorphous selenium
Sectra Microdose – crystalline silicon
GE Essential - CsI
Siemens Mammomat Novation – a-Se
Fuji FCRm – CsBr
Lorad Selenia
Sectra Microdose
GE Essential
Siemens Mammomat Novation
Fuji FCRm
Which of the following systems
does not utilize a grid?
9%
82%
2%
0%
7%
1.
2.
3.
4.
5.
Lorad Selenia
Sectra Microdose
GE DS
Siemens Mammomat Novation
Fuji FCRm
17
Which of the following systems
does not utilize a grid?
1.
2.
3.
4.
5.
Lorad Selenia
Sectra Microdose
GE DS
Siemens Mammomat Novation
Fuji FCRm
Which of the following yields the
overall lowest DQE?
2%
9%
0%
68%
20%
1.
2.
3.
4.
5.
Which of the following systems
does not utilize a grid?
Lorad Selenia
Sectra Microdose
GE Essential
Fischer Senoscan
Fuji FCRm
Explanation
–
The Sectra MicroDose uses a pre- and
post- collimated slot scan system, inherently
reducing the S/P to a level obviating use of
a grid.
Which of the following yields the
overall lowest DQE?
1.
2.
3.
4.
5.
Lorad Selenia
Sectra Microdose
GE Essential
Fischer Senoscan
Fuji FCRm
18
Which of the following yields the
overall lowest DQE?
1.
2.
3.
4.
5.
Lorad Selenia
Sectra Microdose
GE Essential
Fischer Senoscan
Fuji FCRm
References
Bloomquist, A et al, Quality control for digital mammography
in the ACRIN DMIS trial: Part I, Med. Phys. 33 (3), March 206,
p. 719-36.
Lawinski C, et al, Comparative Specifications of Full Field
Digital Mammography Systems, Medicine and Healthcare
Products Regulatory Agency, Report 05037, Crown
Publishing, July 2005
Lazzari, B et al, Physical characteristics of five clinical
systems for digital mammography, Med. Phys. 34 (7), July
2007, p. 2730-2743.
Monnin, P et al, A comparison of the performance of digital
mammography systems, Med. Phys. 34 (3), March 2007, p.
906-914.
Manufacturer-specific documents: Operator’s Manuals, QC
Manuals, white papers, marketing blurbs
RadioGraphics, Vol 10, 1111-1131, 1990
19