Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf ·...
Transcript of Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf ·...
![Page 1: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/1.jpg)
MR-Only Simulation
• Daniel Low, Ph.D.
• UCLA Radiation Oncology
![Page 2: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/2.jpg)
Conflicts of Interest
• Scientific Advisory Board: ViewRay
![Page 3: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/3.jpg)
MRI in RT Departments • Workshop in 2012 November 1, Brigham Women’s Hospital
– U Penn – Wake Forest – Beaumont – U Michigan – Princess Margaret – Hopkins – Duke – U Iowa – Fox Chase – Mayo – Wash U – Harvard – Medical College Wisconsin – Wake Forest – St. Jude – UCLA
MCW
Duke
U Penn U Mich
St. Judes
Princess M Organized by James Balter
![Page 4: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/4.jpg)
Why MR Sim?
• Hypothesis: We are very precise, but not so accurate with radiation therapy
• Precision comes from accurately knowing the dose within the patient
• However, tumor segmentation is often unreliable, so dose is inaccurately optimized
• MR Sim will improve accuracy
![Page 5: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/5.jpg)
NKI: Steenbackkers et al, Red J 64 (2006) 435
No PET
With PET
![Page 6: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/6.jpg)
Sites
Rectal
Prostate
GYN
Breast
McGill: Devic, Med Phys 39 (2012) 6701
![Page 7: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/7.jpg)
Why not BOTH MR and CT Sim?
• Reimbursement for one sim only
• No widespread demonstration of clinical need
• Solution: MR Sim only?
• MR Sim only will increase accuracy, but degrade precision (dose will be less accurately known)
![Page 8: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/8.jpg)
MR Simulation
• MR providing primary or sole 3D imaging information • What does CT give us that we need to replace or
manage? – Bigger bore (85-90 cm vs 70 cm) – Unobtrusive imaging setup (coils) – Flat couches – Spatial integrity – Electron density (from HU) – Reference kV images (e.g. DRRs) – 4D imaging – Lung cancer imaging – Training
![Page 9: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/9.jpg)
Smaller Bore
• 70 cm bore: Modify immobilization and patient setup
• Back to old days with smaller bore
• Modify patient positioning
• Modify immobilization systems
![Page 10: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/10.jpg)
Obtrusive Imaging
• MR requires coils placed near the anatomical sites of interest
• Potential for distorting anatomy
– Floating
– Rigid immobilization
– Built-in coils
![Page 11: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/11.jpg)
Spatial Integrity
• Most modern MRI scanners shim so they are within 2 mm in a ~40 cm diameter sphere
• Multiple sweet spots can be chained together by abutting image acquisitions
![Page 12: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/12.jpg)
Image Series’
• Managing other image acquisitions (functional)
– Develop spatially robust MR acquisition, use analogous to CT in PET/CT
– Register other acquisitions to reference images
![Page 13: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/13.jpg)
Spatial Accuracy
• Need to have validation on per-patient basis that images are spatially robust
– Daily may be insufficient
– Patient-based distortions: chemical and metal
– Machine-based distortions: Permanent markers?
![Page 14: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/14.jpg)
Electron Density • Used for dose calculation
• CT: Essentially is free due to physics of attenuation coefficients of 120 kV and electron density
0.1
1
10
100
1000
10000
Water GrayMatter
WhiteMatter
Liver Fat Bone
T1 (ms)
T2 (ms)
![Page 15: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/15.jpg)
Solutions?
• Water
• Bulk density assignments
• Brain: 2-3% differences heterogeneities on vs off (Ramsey and Oliver, MP 25 (1998) 1928)
• Head Neck: CT vs Water vs 3 levels (bone, air, tissue) bulk assignment. Water = 4%-5%, bulk = 2% (Karotki et al, JACMP 12 (2011) 97)
![Page 16: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/16.jpg)
Prostate
• Lee et al (Green J 66, 203, 2003)
• Water, bulk density, original CT
• 4 patients
• 64 Gy Rx
![Page 17: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/17.jpg)
Results
0
0.5
1
1.5
2
2.5
1 2 3 4
Do
se D
iffe
ren
ce (
%)
Patient Number
Water vs CT
Outside 60%
Inside 70%
Inside 80%
Inside 90% 0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1 2 3 4
Do
se D
iffe
ren
ce (
%)
Patient Number
Bulk vs CT
Outside 60%
Inside 70%
Inside 80%
Inside 90%
![Page 18: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/18.jpg)
Reference Images
• 3D: Possible to use MR directly using mutual information
• 2D: Need projection DRR for comparison – Develop quasi-HU image
– Mostly bone and air contrasts
– Atlas-based mapping of MR images to CT
– UTE imaging for bone imaging
![Page 19: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/19.jpg)
UTE bone imaging ___ patient setup simulation
UCLA
![Page 20: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/20.jpg)
Managing MR Simulation
• Tissue classification via Dixon MR (water-fat imaging) techniques (Helle et al, Proc Int Soc Mag Reson 21 (2013))
![Page 21: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/21.jpg)
4D Imaging (Planar)
• 2-20 fps, low latency
![Page 22: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/22.jpg)
Magnetic Susceptibility
• Electron clouds perturbed by asymmetric molecules and nearby atoms
• Perturbation changes local magnetic field
Paramagnetism Diamagnetism Ferromagnetism
Local field change = positioning error (distortion)
![Page 23: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/23.jpg)
MR Susceptibility
Stanescu et al, Med Phys 39 7185
Segment Bulk Assignment Magnetic Field
Simulation Evaluate
Distortion
B0 orientation 0-3T Cavity Size Ge
Susceptibility = Degree of magnetization of a material in response to applied magnetic field
![Page 24: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/24.jpg)
Results: Air Cavity
1-80 mm diam
Susceptibility: Air Cavity Chemical Shifts: Human Tissues
0
1
2
3
4
0 0.2 0.4 0.6
0.5T Vs 3T
3T
(mm
)
0.5T (mm)
80 mm
1 mm
![Page 25: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/25.jpg)
Improving Spatial Integrity
• Crijns et al, PMB 57 (2012) 1349
• Use only phase-encoding
• Undersample and regularize to reduce imaging time and improve patient stability
• Compare spin-echo to phase encoded sequences Phase encoded sequences were spatially accurate
• Acceleration factor of 4 yielded spatially accurate images
![Page 26: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/26.jpg)
Results
Water/Fat Chemical Shift
Water/Air Susceptibility
![Page 27: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/27.jpg)
Patient Scans
Water/Fat Shifts
![Page 28: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/28.jpg)
Limitations
• 2D acquisition (still need slice selection)
• Improve spatial accuracy in 3rd dimension with strong slice selection gradients
• Inefficient sequences lead to noisy or coarse resolution images
![Page 29: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/29.jpg)
Training
![Page 30: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/30.jpg)
Lun
g C
ance
r?
Pla
nar
x r
ay
Biederer et al Insights Imaging 2012
No
Co
ntrast C
T
T1 3
D G
RE
Fat-
Sat
T2 F
SE
SSFP
T2
![Page 31: Daniel Low, Ph.D. UCLA Radiation Oncologyamos3.aapm.org/abstracts/pdf/77-22638-312436-91687.pdf · 2013. 8. 5. · •Daniel Low, Ph.D. •UCLA Radiation Oncology . Conflicts of Interest](https://reader033.fdocuments.net/reader033/viewer/2022051920/600d19b43af153578f20e131/html5/thumbnails/31.jpg)
Conclusions
• Many steps to get from CT to MR simulation
• My opinion: Biggest unknown challenge is reference image generation
• Unproven potential for improving treatment planning
• Potential for routine functional imaging