The ViewRay, Inc. Renaissance™ James F. Dempsey, Ph.D. C.S.O. ViewRay Inc. Gainesville, Florida
Outline •The clinical problem •Technical rationale •The Renaissance™ •Feasibility Data •Summary
Great progress in optimizing dose delivery to static objects Technology Evolution CT Sim Convolution IMRT Optimization Monte Carlo IMPT etc.
We have perfected the optimization of dose to static objects However…
The Clinical Challenge - Accurately delivery ionizing radiation to the real dynamic patient
4D CT Data from Low et al. Med. Phys. 30(6) (2003) 1254-1263.
Inter-fraction motion studies –few patients, large motions Organ/Tumor
# of Studies
# of Patients
Motion Range [mm]
Inter-fraction Motion Bladder
7
11-30
27 A.P. 4% vol. loss per week 40-80% vol. change
Gynecological Tumors
1
29
<7 Sup. <4 Pos.
Prostate
18
6-55
5.3-20.0 A.P. 1.7-9.9 S.I. 2.0-8.8 Lat.
Rectum
5
11-30
17-76 Dia. Change 6%/week vol. decrease
6-50
1.5-22.0 A.P. 0.35-14.0 S.I. 0.3-5.5 Lat.
Seminal Vesicles
5
Jones and Langen Int. J. Rad. Oncol. Biol. Phys., Vol. 50, No. 1, pp. 265-278, 2001
Intra-fraction motion studies –few patients large motions Organ/Tumor
# of Studies
# of Patients
Motion Range [mm]
Intra-fraction Motion Diaphragm
6
5-30
5-40 Normal Breathing 25-80 Deep Breathing
Kidneys
6
8-100
2-40 Normal Breathing 4-86 Deep Breathing
Liver
5
9-50
7-38 Normal Breathing 10-103 Deep Breathing
Lung Tumors
2
20
5-22 A.P. 0-16 Lat. 1.3-6.5 S.I.
Pancreas
2
36-50
10-30 Normal Breathing 20-80 Deep Breathing
55
No Motion in EPID 0-15 Transient motion with Ciné MRI
Prostate
3
Jones and Langen Int. J. Rad. Oncol. Biol. Phys., Vol. 50, No. 1, pp. 265-278, 2001
Lung Tumor Inter- and IntraFraction Motion Changes All the Time
Hiroki Shirato, Keishiro Suzuki, Gregory C. Sharp, Katsuhisa Fujita, Rikiya Onimaru, Masaharu Fujino, Norio Kato, Yasuhiro Osaka, Rumiko Kinoshita, Hiroshi Taguchi et al. Int J Radiat Oncol Biol Phys. 2006 Mar 15;64(4):1229-36.
Real-Time 3D Image-Guidance
Intra-fraction motion occurs continuously -from the base of the tongue to bottom of the pelvisreal-time imaging is the only comprehensive answer
Intra-fraction Organ Motion Example Rectal: Gas Distention
In 1999 Padhani et al. scanned 54 prostate cancer patients in axial plane every 10
second for 7 minutes Padhani et al. Int. J. Rad. Oncol. Biol. Phys., Vol. 44(3) pp. 525–533, 1999 Ghilezan et al. Int J Radiat Oncol Biol Phys. 2005 Jun 1;62(2):406-17.
Intra-fraction Organ Motion Example Rectal: Gas Distention > 0.5 cm Prostate Motion for 20-80 seconds observed in 16% of patients
No considerable motion in 1/2 16.7% (9/54) had prostate move > 5mm median prostate AP displacement was anterior by 4.2 Lasting 10-80s w/ mean of 20s What would the impact on TCP be?
Back-of-the-envelope: Loss of TCP from Prostate Motion TCP Model of Stavrev et al. (Phys. Med. Biol. 50 (2005) 3053–3061)
– α=0.14 [Gy-1] =0.04 [Gy-2] – λ=0.12 [days-1] cell repopulation – τ=0.576 [days] sub-lethal damage repair time
Valid for different dose/time Monte Carlo 5K cases 16.4% chance of X% dose error in fx X = 10,20,30,40,50% TCP @ 5yrs
Adaptive Therapy? Onboard volumetric imaging is here and it allows for •Currently: Takes snapshots before or after therapy & shifting the patient position •Preferably: Automated IMRT re-optimization
A great advance for radiotherapy, but Current technology has no ability to account for intra-fraction motions!
Intra-fraction Motion is Observed in During Cone-Beam CT Acquisition Lung breathing artifacts are clearly evident Rectal gas artifacts seen in prostate for every 1 of 6 cases See Smitsmans et al. Int J Rad Oncol Biol Phys 63(4):975-984
Looking down the CBCT
Real-Time X-Ray based IGRT? CT imaging systems Are currently slow ~1 min. per volume Provide extra dose to the patient Real-time: 1 CT/sec over 5 min. w/ 0.5 cGy/CT = 150 cGy extra! Requires fast moving parts Cone-beam at 1 RPM Multi-Slice CT systems Fast ~0.5 seconds/ image, but small field of view
Why Not MRI? No moving parts! Used for Simulation Very, very fast volume acquisition! Parallel or dynamic MRI No ionizing radiation dose to the patient! MRI can image metabolic & physiologic information
MRI + Linac System = Conflict Mr. Green from Varian Med. Sys. filed patent in 1997 Extensive combinations of linac and MRI
Conceptual System Announced in 2001 by Utrecht University in the Netherlands 6MV Linac +1.5 Tesla MR Simultaneous imaging and radiotherapy will NOT be possible with their device Treating through the device ~20 cm of Al Technically Feasible? Economically Feasible?
MRI vs. Linac The magnetic field will shut off the Linac The Linac RF can destroy delicate circuitry & ruin images
The
TM Renaissance
System 1000
Preliminary Specifications •Superconducting Open 0.3 Telsa MRI w/ 50 cm FOV & 80 cm bore •3 x 13 KCi sources with 750 cGy/min. @ 1 m and double focused MLC
•IMRT or Conformal photon beam therapy •Supercomputing grid for fast •Monte Carlo Simulation including magnetic field •Deformable Image Registration •IMRT Optimization •Parallel MRI Reconstruction
Why Low Field MRI? Low field MRI is a must for radiation therapy because: 1) High field causes a loss of spatial integrity Magnetic Susceptibility artifacts due to the patient scale with Bo field strength e.g. 1 cm distortion at 3T => 1 mm distortion at 0.3T
2) High field ruins the dose distribution see next slide
See Petersch et al. Radiotherapy and Oncology 71 (2004) 55–64 0.3 T -> 3.24 mm max distortion 1.5 T -> 16.2 mm max distortion
Physics of Electron Transport in MRI CSDA electrons in B field Lorentz Force causes a force perpendicular to the magnetic field direction This causes the electrons to gyrate in a circle or spiral if loosing energy Competition between large-angle electron scattering and the radius of gyration In a 0.3 Tesla field the radius of gyration for a 1 Mev electron in vacuum is 1.3 cm In a 1.5 Tesla field the radius of gyration for a 1 Mev electron in vacuum is 3.4 mm
In theory the electron will radiate synchrotron radiation but this is << eV/cm See Beliajew Med. Phys. 20(4) 1993 11711179 And Jette Med. Phys. 27(8) 2000 1705-1716
Photon Beam Dose Distortion @ 1.5 T Significant distortion of the dose in water at 1.5 Tesla & 6MV Electron Return Effect
Raaysmaker et al. Phys. Med. Biol. 49 (2004) 4109–4118 Raaijmakers et al. Phys. Med. Biol. 50 (2005) 1363–1376
60Co
+ Low-Field MRI @ 0.3 Tesla in Tissue (1g/cc)
MC shows Essentially no distortion in tissue or water MFP for large angle collisions of secondary electrons much shorter than radius of gyration
60Co
+ Low-Field MRI @ 0.3 Tesla in Lung (0.2 g/cc)
MC shows very small distortion in lung density material
60Co
+ Low-Field MRI @ 0.3 Tesla in Air (0.002 g/cc)
MC shows sizable distortion only in air cavities only hot spots at interface are greatly diminished
MRI Improves 60Co IMRT Electron Contamination is Swept Away MRI Sweeps Away the Contamination Electrons Even a low-field Open MRI will provide enough field strength to sweep contamination electrons In a 0.3 Tesla field the radius of curvature for a 1 Mev electron in vacuum is 1.3 cm Contamination electrons cannot reach the patient: lower skin dose to patient Can be modeled by Monte Carlo Simulation See paper for measurements of sweeping effect: Jursinic and Mackie Phys. Med. Biol. 41 (1996) 1499–1509.
Elimination of Contamination Electrons
Electrons are shown in blue/white Photons are shown in pink
How to Make MRI Fast @ Low Field:Parallel MRI (pMRI) Current MRI scanners already operate at the limits of potential imaging speed based on rapidly switched gradient systems (for safety concerns). Huge advances from pMRI
– Commercially, up to 32 independent receiver channels available which theoretically allows order-of-magnitude increased image acquisition speed
Sodickson et al. Acad Radiol. 2005 May;12(5):626-35.
What About Signal? Low field MRI is a must for radiation therapy 1.5T => 0.3T Factor of 5 loss of signal But, 1mm voxel => 3 mm voxel gives 27 times more signal still 5.4 times more
vs
Low Field MRI for Simulation & Planning Examples of 0.2 T Open MRI Simulation Data of Lung & Prostate Cancer Patients
Real-time MRI: Lung Coronal & Sagittal 2D MRI taken every 0.5 seconds on an existing 0.2 T open MRI Benefits: Capture 4D target every day Gate therapy on motion of soft tissues
Real-time MRI Coronal & Sagittal 2D MRI taken every 0.5 seconds on an existing 0.2 T open MRI Benefits: Capture 4D target every day Gate therapy on motion of soft tissues
Real-time MRI Coronal 2D MRI taken every 0.5 seconds on an existing 0.2 T open MRI Benefits: Capture 4D target every day Observe effects like blood flow, coughing, swallowing, voluntary motion, IMRT
aliasing w/ motion, etc.
What else can MRI currently bring to the table?! MRI can provide Better soft tissue contrast T1 T2 Proton density
Bold Perfusion imaging Spectroscopy
What else will MRI bring to the table in the future?! Exciting MRI contrasting agents that can provide “ nuclear medicine” -like metabolic information are being developed Hyperpolarized liquids Liposome-based agents
Come for the organ motion, stay for the metabolic imaging!
Why -Ray IMRT? Because it works!!! High quality optimization enables gamma-ray IMRT
40 seconds to optimize on single PC Compatible w/ MRI 1.5 cm =diameter 60Co source 300R/min. @ 1 meter MLC @ 60 cm 7 beam plan Targets to 73.8 and 54 Gy Spare tissue, saliva glands, cord, brain stem, and mandible
Head&Neck Case: DVHs
Targets w/ >95% Vol. coverage <12% hot spot for high dose target Sparing for 3 out of 4 saliva glands <50% vol. @ 30 Gy <3% Tissue > 50 Gy Cord, brain stem, and mandible below tolerance
Renaissance Goes “ Toe-to-Toe”with the Best
a)
6MV
Co60
6MV
Co60
71 beams
71
7
7
beams
beams
beams
b)
c)
d)
By Every Measure Co60 Makes Great Plans
a)
b) 6MVsolid vs 60Codashed 7 beams
6MV 71solid, 11dashed, 5dotted
a )
b )
c )
d )
Prostate - Dose Dist. a)
b)
6MV
6MV
71
7
beams
beams
Co60
Co60
71
7
beams c)
beams d)
Prostate -DVHs 6MVsolid vs
a)
60Codashed
6MV 71solid, 11dashed, 5dotted
7 beams
b) a)
b)
c)
d)
Is Cobalt a Problem? 60Co
is undoubtedly the best isotope for external beam therapy Cobalt is a ferromagnetic metal Ferromagnetic materials magnify magnetic fields Magnetic field inhomogeneities can destroy the performance of the MRI How big is this effect?!
No! Co Has Negligible Effect on MRI! Consider a small 1.5 cm dia. sphere of cobalt in a uniform 0.3T field 1 m away from a 70 cm field of view. Cobalt acts like a soft ferromagnetic material. The magnitude of the magnetic field can be found exactly by solving Poisson’ s Eqn. for the magnetic potential Inside the sphere & on its surface the field is 0.9 T The excess magnetic field falls off as a dipole, i.e., with 1/r3
• Less than chemical shift
Where the cobalt induced field meets the MRI FOV is already ~ 2ppm and rapidly dropping!!!
Can We Compute Dose Without CT Densities ?
Conformal Lung treatment plan: take CT data & reduce to 3 values: lung; bone; and soft tissue; having 0.15,
Computing Dose Without CT
DVH overlay of full CT calc. and 3 density calc. No observable difference in the DVHs
Computing Dose Without CT
We just need to know where the air, lung, soft tissue, and bone are Dose Difference < +34 cGy or 0.5% everywhere By the way, you can use this with CBCT… CBCT…
Can We Differentiate Tissues in MRI?
T2
T1
Yes, Using the information in both T1 & T2 pretreatment MRIs we can differentiate bone from air
Differentiating Tissues in MRI
The rectal gas and air selected over pelvis & femurs
Roadmap •The Generation of the Renaissance •Gen-1 - daily reoptimization & dose recording •Gen-2 - closed loop beam-by-beam reoptimization •Gen-3 - real-time reoptimization driven tracking •Metabolic Imaging TM
Summary & Outlook Viewray, Inc. Formed w/ experienced management Patent pending Feasibility Studies Completed Forming scientific board of advisors Design team established Strong Corporate Partners with experience in: whole body MRI, Cobalt Therapy, MLC systems, control systems, gantry & couch design Seeking strong clinical institutions & strategic partners for collaboration on ViewRay development for Adaptive treatment planning algorithms in HPC system MRI metabolic imaging Deformable image registration
pMRI development Metabolic imaging
Collaborators James F. Dempsey, Ph.D.1, Andrew W. Beavis, Ph.D.2, Benoit Dionne, M.S.3, Jeffrey F. Fitzsimmons, Ph.D.4, Alireza Haghighat, Ph.D.3, Jonathan G. Li, Ph.D.1, Daniel A. Low, Ph.D.5, Sasa Mutic, M.S.5, Jatinder R. Palta, Ph.D.1, H. Edwin Romeijn, Ph.D.6, and Glenn E. Sjoden, Ph.D.3 1)Department of Radiation Oncology, University of Florida, Gainesville, Florida 32610, U.S.A. 2) Department of Medical Physics, Hull and East Yorkshire NHS Trust and Institute of Clinical Bio-Sciences, University of Hull. Princess Royal Hospital, Saltshouse Road, Kingston Upon Hull, HU8 9HE, England, UK. And Visiting Professor, Faulty of Health and Well Being, Sheffield-Hallam University, Sheffield, UK 3)Department of Nuclear and Radiological Engineering, University of Florida, Gainesville, Florida 32611 4)Department of Radiology, University of Florida, Gainesville, Florida 32610, U.S.A. 5)Department of Radiation Oncology, Washington University, St. Louis, Missouri 63110, U.S.A. 6) Department of Industrial and Systems Engineering, University of Florida, Gainesville, Florida 32611, U.S.A.
Collaborators Robert J. Amdur Benoit Dionne Jeffrey F. Fitzsimmons Alireza Haghigat Jonathan G. Li Chihray Liu Daniel A. Low Sasa Mutic Kenneth Olivier Jatinder R. Palta H. Edwin Romeijn Glenn E. Sjoden Ilona Schmalfuss John Ziegert Robert Zlotecki
UF Rad. Onc. UF NRE UF Radiology UF NRE UF Rad. Onc. UF Rad. Onc. Washington Univ. Washington Univ. UF Rad. Onc. UF Rad. Onc. UF ISE UF NRE UF Radiology UF MAE UF Rad. Onc.
