7/13/2015
RADIATION ONCOL OGY &
MOLECUL AR RADIATION SCIENCES
Development of Intra-Fraction Soft Tissue
Monitoring with Ultrasound Imaging
Presented by: Kai Ding, Ph.D.
Department of Radiation Oncology, Johns Hopkins University, Baltimore, MD
July 13, 2015
AAPM 2015, Anaheim, CA – July 12-16, 2015
Disclosure
• This research is supported in part by Elekta
and NCI CA161613
Acknowledgement
HUG (Hopkins Ultrasound Group)
• Radiation Oncology
– John Wong, Lin Su, Yin Zhang, Sook Kien Ng,
Junghoon Lee, Ted Hooker, Joseph Herman,
Harry Quon, Phuoc Tran, Danny Song
• Engineering
– Iulian Iordachita, H. Tutkun Sen, Peter
Kazanzides, Muyinatu A. Lediju Bell
• Radiology
– Jinyuan Zhou, Chen Yang
1
7/13/2015
Background
Seed based
snapshots
Non ionizing
Seed based
Real time
Non ionizing
Non invasive
Real time
Background
• US is portable, non-invasive, and non-ionizing
modality for daily setup and real time monitoring
US challenge 101
Who is going to hold the
probe during radiation?
Liver
Prostate
Pancreas
2
7/13/2015
US challenge 102
Can the probe be placed
in a repeatable way?
Liver
Prostate
Pancreas
Two solutions
Passive arm (arm
bridge system)
+Small size
+Easy docking
-User dependent
Active robotic arm
(UR5)
+Less user dependent
+Feedback loop
-Complex docking
Arm bridge system (ABS)
3
7/13/2015
Arm bridge system (ABS)
Arm bridge system (ABS)
Probe impact on planning
4
7/13/2015
Probe impact on planning
Step n shoot
VMAT
Clinical
Probe impact on planning
PTV
Clinical
Probe ss
P(1)
Probe vmat
P(2)
P(3)
D95(dose to 95% of the PTV) [cGy]
3168±121
3151±78
0.552
3195±80
0.459
0.009
D_mean (mean dose to PTV) [cGy]
3781±63
3709±63
0.004
3749±82
0.383
0.133
V33 (ratio of PTV receiving 33 Gy) [%]
92.8±2.0
91.9±1.4
0.187
93.3±1.3
0.600
0.002
volume of 33Gy + [cm^3]
83.4±29.7
80.4±31.3
0.155
82.2±30.5
0.747
0.242
CI (conformal index)
1.08±0.07
1.03±0.08
0.070
1.06±0.09
0.369
0.222
D5 (dose to 5% of the PTV) [cGy]
4122±124
4020±96
0.020
4043±118
0.387
0.610
HI ( homogeneity index)
0.25±0.06
0.23±0.03
0.130
0.23±0.03
0.389
0.458
OAR
Clinical
Probe ss
P(1)
Probe vmat
P(2)
P (3)
duo-V15 (cc)
5.66±2.00
6.01±2.17
0.172
5.24±2.13
0.371
0.037
sto-V15 (cc)
6.00±2.54
6.12±2.94
0.825
4.92±2.45
0.139
0.035
bowel-V15 (cc)
6.09±2.05
5.78±2.00
0.271
4.64±1.99
0.005
0.004
liver V12 (%)
6.18±5.01
7.02±6.91
0.312
6.96±7.06
0.467
0.959
kidney V12 (%)
6.36±6.65
7.08±7.07
0.020
7.83±7.55
0.112
0.326
stomach V12 (%)
7.51±4.26
7.01±3.67
0.439
5.67±3.30
0.065
0.027
Real time monitoring
5
7/13/2015
Repeat probe position
Arm bridge system (ABS)
Free breathing monitoring
Liver
6
7/13/2015
Active breath hold monitoring
Pancreas
US Challenge 102
Can the probe be placed
in a repeatable way?
Liver
Prostate
Pancreas
US challenge 102
When the US probe is placed on the
patient, the target organ deforms
US Probe
Simulation Room CT
Treatment Room LINAC
For accurate radiation
treatment, same soft tissue
deformations should be
created between
Simulation and Treatment
7
7/13/2015
Active robotic arm
• Less user dependent: can record the position and force of the probe and
repeat probe placement
• Feedback: can fine adjust the position and force based on the acquired US
image
• Complex docking: docking system needs to be designed for quick clinical
setup
Experiment
Goal :
1. To place the target organ to the simulation day position
2. To create the same soft tissue deformation between sessions
Experiments :
3 metal markers are implanted into a
canine kidney to represent a tumor
US
Probe
Implanted markers are localized in the US
and CT images taken in the experiment
Marker positions in the CT images
are used as the ground truth
Experiment
8
7/13/2015
Experiment
Placement of US
probe with robot
Bring the US probe directly to the simulation day position
(Hard Virtual Fixture)
Record Probe position,
take US and CBCT
inter-fraction images
Activate the soft virtual springs
Clinician searches for the
reference US image
NO
𝑷𝑯𝒂𝒓𝒅 𝑽𝑭
Shift the couch based
on the marker positon
differences of US images
Take a CBCT and compare
with the reference CT
US image Found?
YES
Shift the couch based
on the marker positon
differences of US images
Record Probe position,
take US and CBCT
inter-fraction images
𝑷𝑺𝒐𝒇𝒕 𝑽𝑭
Experiment
Claim :
The soft virtual fixture helps to create the same
soft tissue deformation
For each interfraction day :
Marker position differences between Hard VF
and Reference US images & Soft VF and
Reference US images are compared
Reference
Hard VF
Soft VF
Experiment
Compared to the alignment to using CBCT of spine only:
With the US image feedback, the setup accuracy is improved by 42 %
With the soft virtual fixture algorithm, it is improved by 61%
9
7/13/2015
Summary
Passive arm (arm
bridge system)
+Small size
+Easy docking
-User dependent
Active robotic arm
(UR5)
+Less user dependent
+Feedback loop
-Complex docking
Questions
10