Recent progress on GPU-based Monte Carlo Simulations for Radiation Therapy Xun Jia, Ph.D. Xun.Jia@UTSouthwestern.edu Radiation Oncology
Outline • Recent progress • Two packages • Considerations • Conclusion Radiation Oncology
Outline • Recent updates Radiation Oncology
GPU 4-16 cores >1000 cores Radiation Oncology
GPU Processing Memory Price Clock Rate Core power (MB) ($) (MHz) (GFLOPS) Geforce GTX TITAN black (Feb 2014) • 5120 (SP) • 999 • 2880 • 889 • 6144 • 1706 (DP) Geforce GTX (Mar 2017) • 3584 • 1417 • 11264 • 10609 (SP) • 699 1080 Ti • 332 (DP) Radiation Oncology
GPU-MC project at UTSW 2009 2011 2012 2014 2015 2016 gDPM gCTD gPMC goMC goCMC goMicroMC gMCDRR gBMC • Particle types: photon, electron, proton, carbon ion, free radical… • Energy ranges: eV keV MeV GeV • Spatial scales: nm (DNA level) m (human level) • Clinical applications: external beam therapy, brachytherapy Radiation Oncology
Particle therapy • gPMC goPMC • Race condition Qin et. al. PMB, 61, 7437 (2016) Radiation Oncology
Particle therapy • goCMC • CSDA • Energy straggling and angular deflection • Nuclear interaction • Considering only interactions with H, C, O, and Ca • Tabulated data prepared with Geant4 • Secondary neutral particles neglected • Simulation time of 10 7 C 12 : 11~162 sec (100~400 MeV/u) Qin et. al. PMB, 62, 3628(2017) Radiation Oncology
Particle therapy • Biological dose calculation with RMF model Qin et. al. To appear in Red Journal (2017) Radiation Oncology
Particle therapy • Biological inverse optimization • Full GPU-MC based biological optimization Qin et. al. To appear in Red Journal (2017) Radiation Oncology
Particle therapy • Front interface in Eclipse Qin et. al. To appear in Red Journal (2017) Radiation Oncology
Geometry modeling • Voxelized geometry Quadratic geometry • Stored in a tree structure • Two key geometry functions • Time vs memory type Chi et. al., PMB 61, 5851 (2016) Radiation Oncology
Geometry modeling • Memory-speed tradeoff 1 • An auxiliary array of body index in texture memory -1� 0� � � � � � � � • Time vs memory size Chi et. al., PMB 61, 5851 (2016) Radiation Oncology
Geometry modeling • Applications • PET detector simulation r� =� 0� 5.5� cm� Radiation Oncology
Microscopic MC • gMicroMC s Ionizing� radia on� Time� 0 H 2 O� Ioniza on� excita on� Physical� stage� 10 -15� s� + H 2 O +� e -� H 2 O*� Simula on� me:� dissocia on� seconds� to� minutes� Physico-chemical� solva on� dissocia on� relaxa on� stage� e ↓ auto-ioniza on� aq 10 -12� H 3 O +� s� · OH� H · � H 2� ↑ diffu i o n� Simula on� me:� � Chemical� stage� − � chemical� reac on� up� to� hours� or� days� e ↓ aq H 3 O +� 10 -6� · OH� H · � H 2� H 2 O 2� OH -� s� ↑ − � Tian et. al., PMB 62, 3081 (2017) Radiation Oncology
Microscopic MC • Chemistry stage • Step-by-step diffusion reaction model • Brownian bridge considered • Complexity due to chemical interactions • Particle binning with reaction radius • Search reactant within neighbors Simulation time (s) N Speed- up Geant4- gMicroMC DNA 750 keV 101829 102865.4 599.2 171.1 electron 5MeV proton 56122 96446.5 489.0 197.2 Tian et. al., PMB 62, 3081 (2017) Radiation Oncology
Microscopic MC 3 10 gMicroMC ICRU 16 ICRU 37 2 /g) 2 10 Sopping power (MeV cm 1 10 0 10 1 2 3 4 5 6 7 8 10 10 10 10 10 10 10 10 Energy [eV] Tian et. al., PMB 62, 3081 (2017) Radiation Oncology
Outline • Two packages Radiation Oncology
Two packages • goMC • Coupled photon/electron transport with quadratic/voxelized geometry Dense Water (2.0 g/cm 3 ) Bone (1.85 g/cm 3 ) 6x photon Water (1.0 g/cm 3 ) Lung (0.3 g/cm 3 ) Radiation Oncology
Two packages • gMicroMC Radiation Oncology
Outline • Considerations Radiation Oncology
Considerations • MC in the rapid (GPU) parallelization era • New algorithms vs Embarrassing parallelization • Speed-memory tradeoff 1 -1� 0� � � � � � � � Radiation Oncology
Considerations • MC in the rapid (GPU) parallelization era • Single vs double precision • Cross platform • OpenCL gDPM (s) goMC (s) No. of Ratio goMC (s) Beam Phantom Nvidia GeForce Nvidia GeForce particles goMC/gDPM NVidia Intel i7-3770 Intel i7-3770 No. of AMD AMD GTX TITAN GTX TITAN Beam particles Phantom GeForce CPU CPU Radeon Radeon HD 3.7 ± 0.2 4.3 ± 0.1 Water 1.16 15MeV GTX (4 cores, 8 (single thread) 5 × 10 6 R9 290x 7570 4.4 ± 0.1 4.9 ± 0.1 TITAN threads) electron Slab 1.11 4.3 ± 0.1 4.7 ± 0.2 123.9 ± 1.4 51.7 ± 1.7 213.4 ± 5.2 Water 15MeV 35.6 ± 0.2 36.9 ± 0.0 Water 1.04 6 5 × 10 4.9 ± 0.1 5.3 ± 0.1 142.4 ± 0.8 59.2 ± 0.9 22 4.5 ± 7.6 6MV electron Slab 5 × 10 8 44.1 ± 0.1 50.2 ± 0.2 Slab 1.14 36.9 ± 0.0 31.4 ± 0.1 1441.0 ± 3.2 471.4 ± 4.0 2139.1 ± 2.4 Water photon 43.0 ± 0.0 48.6 ± 0.2 Half-Slab 1.13 50.2 ± 0.2 36.3 ± 0.3 1766.6 ± 0.7 51 1.6 ± 9.4 2943.4 ± 17.9 6MV Slab 8 5 × 10 photon Half- 48.6 ± 0.2 36.0 ± 0.2 1781.4 ± 17.8 521.1± 6.8 2981.5 ± 10.3 Slab Radiation Oncology
Outline • Conclusion Radiation Oncology
Conclusion • Continuous development of GPU-based MC • New physics regimes • New capabilities • New applications • Two packages open for testing and collaborations • How to best use GPU’s power in an MC problem? Radiation Oncology
Conclusion • Speed is … • Speed • Accuracy • Big data Radiation Oncology
Acknowledgement • UTSW team • Steve B. Jiang • Nan Qin • Min-Yu Tsai • Zhen Tian • Yujie Chi • … • Collaborators • Harald Paganetti and team @ MGH • Katia Parodi and team @ LMU • Funding support Radiation Oncology
Recommend
More recommend
