第5讲-计划系统与剂量计算
- 1. 计划系统与剂量计算 Guanghua Yan, Ph.D., DABR University of Florida ufyan@yahoo.com 04/06/2015
- 2. 提纲计划系统在放疗过程中的地位 Linac 物理模型 计划系统剂量计算算法 基于测量 基于模型 Convolution/Superposition Monte Carlo 计划系统commissioning Commissioning的过程 测量数据2
- 3. 基本放疗流程Courtesy of Dr. Jatinder Palta1.咨询 2.摆位与定位 3.成像(CT, MRI) 4.勾画(target, OAR) 5.计划制定 6.计划验证 7.计划执行 两大步骤-制定计划和执行计划CT simTPSLinac3
- 4. 计划制定放疗计划的制定依赖于计划系统 (Treatment planning system) 针对具体的病人 医生定义靶区和处方 剂量师产生一组照射野 这组照射野的剂量分布（isodose lines）应该 Maximize tumor control Minimize normal tissue complication4
- 5. 计划执行计划执行-直线加速器 (Linear Accelerator) 拟定计划从计划系统传输到 RV (Record and Verify) 系统，再传到Linac Linac 按照照射野参数逐一执行 目标 “病人受到的剂量分布=计划的剂量分布” or “delivered dose = planned dose”；几个关键： 医生勾画精度 病人摆位准确性 病人有否移动 计划系统准确性 执行系统可靠性5
- 6. 计划系统 (TPS)计划系统 – virtual machine CT image as virtual patient 计划系统模拟机器(Linac)的行为 利用虚拟的机器优化机器参数达到治疗目的 计划系统有很多功能，包括图像处理、图像融合、靶区勾画、剂量计算、计划设计、优化计划等 本次课的重点是剂量计算，我们讨论计划系统的 物理模型 (以Pinnacle为例) 剂量计算算法 (侧重于Convolution/Superposition) 计划系统commissioning 6
- 7. 计划系统物理模型 (Physics, beam model)7
- 8. 直线加速器 (LINAC)-高能电子的产生Courtesy - EB Podgorsak, Radiation oncology phyiscs: a handbook213564Accelerating waveguide Electron gun Microwave power source Modulator Beam bending Gantry head从5输出的电子束需具备哪些属性？8
- 9. 直线加速器 (续)Linac的目的是产生高能电子束，同时具备这些属性 a. High energy b. Minimum energy spread c. Focused (细) d. Straight e. Centered 这些跟计划系统物理模型有关吗？tilteddisplacedidealEnergy spectrum9
- 10. Treatment head – 光子的产生高能电子束撞击X-Ray Target (High-Z 金属)，产生高能光子 Bremsstrahlung process Forward-peaked photon distribution especially for high energy electron beam Energy spectrumBremsstrahlung processForward peaked photon distributionPhoton beam energy spectrum10
- 11. Treatment head – beam modifyingPatient independent X-Ray target，简称target，电子到光子 Primary collimator – 50 cm 直径 Flattening filter Ion chamber – beam监控 Patient dependent Beam shaping device-改变beam的形状 Secondary collimator (Jaw),两对垂直的铅门 (X Jaw, Y Jaw) Multileaf collimator (MLC) 11
- 12. Treatment head – photon sourceFocal source High-Z target，可以想象成一个”isotropical”的点源 (point source) Extra-focal source due to scatter Primary collimator Flattening filter Ion chamber (minimal) 比重 ~12% Backscatter From Jaw to Ion chamber 比重 1~2% A Ahnesjo et al, PMB,44, R9912
- 13. Dose from primary beam Primary dose Phantom scatter dose Dose from scatter beam Charged particle contamination dose Head scatter doseFour dose categories from Linac13
- 14. Ahnesjo et al., PMB 199914
- 15. 计划系统剂量计算算法15
- 16. Measurement-based algorithms 基于测量 基于经验的手工计算，采集大量的数据，利用插值计算dose in water，再修正 (apply corrections for pt contour & heterogeneity) Good for simple geometries Model-based algorithms 基于模型 Convolution/superposition Monte Carlo光子剂量计算算法分类16
- 17. MU: monitor unit; TD: tumor dose; ID: isocenter dose %DD: PDD; TMR: tissue maximum ratio Sc/Sp – head/phantom scatter factor Can be used for spot-check on TPS Used to calculate MU for conformal tx Not accurate for small/irregular field size基于测量的MU 计算SSD setupSAD setupKhan’s book17
- 18. 基于模型的剂量计算 Convolution/Superposition methods （卷积/叠加算法）18
- 19. 几个重要的物理概念几个重要的与剂量计算相关的物理概念 KERMA TERMA Absorbed Dose Photon energy fluence Ψ Coefficients Mass energy transfer coefficient utr/ρ Mass energy absorption coefficient uab/ρ Mass attenuation coefficient u/ρ Khan, The Physics of Radiation Therapy19
- 20. TERMATERMA: Total energy released in the medium, defined as total radiant energy released per unit mass by primary photons in a medium total energy (Energy of electron + energy of scatter photon) 只考虑primary photons,不考虑从介质中其它地方scatter过来的photons 20
- 21. Absorbed doseAbsorbed dose, or simply dose, is the quotient dE/dm, where dE is the mean energy imparted by ionizing radiation to material of mass dm. 简而言之，dose = 单位质量吸收的能量 TERMA和Dose的单位-MeV/g21
- 22. Energy fluenceCourtesy - EB Podgorsak, Radiation oncology phyiscs: a handbook假想有个很小的小球，其截面积是dA，energy fluence刻画有多少能量穿过这个截面; 单位 – MeV/cm222
- 23. Coefficients 系数不是很严格的理解，系数就是比例,单位 cm2/g Attenuation coefficients - u/ρ 衰减系数-多少比例的粒子会产生反应 Energy transfer coefficients - utr/ρ 能量传递-多少比例的能量会传递给介质 Energy absorption coefficients - uen/ρ 能量吸收-多少比例的能量会被介质吸收23
- 24. TERMA计算TERMA: Total energy released, by primary photons where Ψp is energy fluence of primary photons at a point, and 𝛍/𝛒 is mean mass attenuation coefficient Note: Primary photons only! Mass attenuation coefficient 24
- 25. 从TERMA 到 Dose如果光子产生了反应，它“释放”掉所有的能量到介质中 E.g., Compton Scatter 部分能量locally被吸收 (Primary dose, D(r)), 部分能量被scattered photons带走，在介质中其它地方被吸收 (scattered dose, D(r’)); 极少部分会逃离介质 (E(x)); 能量沉积分配？- KernelD(r’’)Primary photon fluenceD(r’)D(r)E(x)water25
- 26. Energy deposition kernel 能量沉积分布 例子，假设某个点释放的TERMA是T,那么同一个水平面，离这个点100 mm处沉积的剂量是Tx10-4 Dose(r’) = T(r) x K(r,r’) 3D, rotational symmetric Kernel 包含了phantom scatter dose! 所以算TERMA时，只需考虑primary photons! 26
- 27. 剂量计算-两步第一步，计算介质中每个点primary photon释放的能量TERMA 核心-计算每个点的primary photon energy fluence Ψp Primary photon怎么产生的？ 第二步，计算每个点的TERMA在介质中能量的沉积 核心-energy deposition kernel 通过蒙卡(Monte Carlo) （近似）在整个介质中，kernel不变 Mackie et al 发表kernel for 6 MV 27
- 28. Treatment head – photon sourceFocal source High-Z target，可以想象成一个”isotropical”的点源 (point source) Extra-focal source due to scatter Primary collimator Flattening filter Ion chamber (minimal) 比重 ~12% Backscatter From Jaw to Ion chamber 比重 1~2% A Ahnesjo et al, PMB,44, R9928
- 29. Primary photon energy fluence 计算1从photon源的角度 Ray-tracing 从上到下 Focal source –点源； extra-focal source (e.g., 二维高斯函数) 离散化成一个个点源 A Ahnesjo et al, PMB,44, R99源介质29
- 30. Primary photon energy fluence 计算2从介质的角度 Back projection 从下往上计算有多少可见源 (visible source) 源可能被Jaw或者MLC挡住 Jaw/MLCJaw/MLCTERMA (可见源积分)Extra focal source due to Flattening filter30点源
- 31. Convolution –卷积计算任给一点 r’的剂量 计算TERMA at any point r: T(r) e.g. T(r1), T(r2) , T(r3) r1贡献给r’的剂量 T(r1) x k(r1-r’) r2贡献给r’的剂量 T(r2) x k(r2-r’) r3贡献给r’的剂量 T(r3) x k(r3-r’) Dose at location r’ D(r’) = T(r1) x k(r1-r’) + T(r2) x k(r2-r’) + …T(r1): TERMA at location r1 T(r2): TERMA at location r2 T(r3): TERMA at location r3卷积D(r’)T(r1)T(r2)T(r3)源31为什么转换成卷积？-利用傅立叶变化理论加速快速
- 32. Heterogeneity correction人不都是水-有骨头有肉，有心有肺 CT density heterogeneity Needs correction for dose calculation Two directions Longitudinal direction (沿着Primary beam) Ray tracing算TERMA时，考虑附加的attenuation Lateral direction (和primary beam垂直方向) 算T(r)对r’的剂量贡献时， 不用 T(r) x K(r-r’) 用 T(r) x K(reff), reff-effective path length, 考虑电子密度 相当于挤压或者拉伸Energy deposition kernel r’r32
- 33. 33
- 34. 34
- 35. Convolution/superposition剂量计算小结1. treatment head - 已知光子源的分布 Focal source & extra focal source 又叫beam model, source model, etc 2. Ray-trace – 从源到病人体内每个点 计算病人体内每个点的TERMA 在Ray线上考虑附加的衰减 3. 计算每个点的TERMA在体内所有点的剂量沉淀 基于dose deposition kernel 考虑密度变化，拉长或者挤压kernel 计算速度？- collapsed cone convolution r’r源35
- 36. Collapsed Cone Convolution (CCC)假设病人体内某点O的TERMA是T(O) O贡献剂量到三维空间的每个点 一个个点去计算，time consuming Collapsed cone convolution (CCC) 以O为圆心 把三维空间分成一个个圆锥 对每个圆锥 只有中轴线上的点收到剂量 圆锥内的剂量都挤压在中轴线上 节省时间 对病人体内每个点，都做这个过程 省时 精度影响很小 36
- 37. Monte Carlo 算法概念上来说，MC算法更直截了当 (brute force) 我们可以跟踪每个撞击Target的电子，按照各种物理反应事件的概率（coefficients），掷骰子，去模拟各种反应，同时记录每个点的剂量Target BackingPrimary CollimatorFlattening Filter PSD PlaneTargetPatientJawsWedgese-e-ee-e-+BremsstrahlungBremsstrahlungComptonComptonPairAnnihilationContinuous slowing downJaws37
- 38. Convolution/superposition 与 Monte Carlo比较C/S Calculate dose from physics principles; intuitive Accuracy has been validated against MC State-of-the-art algorithm Monte Carlo Most accurate under all situations (in theory) Model verification? Time consuming 38
- 39. 计划系统 commissioning Pinnacle beam model Pinnacle commissioning39
- 40. Philips PinnacleTM40
- 41. Pinnacle beam model的组成Focal source-”点”源 (不完全是个点)-finite size Extra-focal source (primary collimator & Flattening Filter) 2D Gaussian function (width & height) Beam spectrum Note beam spectrum changes with position Beam quality softening factor Beam intensity variation across the field Due to preferential attenuation of FF Inverse cone to modify beam intensity Transmission through Jaw and MLC Multi-leaf collimator 对于IMRT非常关键 Rounded leaf-end, Tongue & groove, inter & intra leaf transmission 总而言之-beam model用来描述光子源的分布，最后用于计算TERMA! 41
- 42. Pinnacle beam model commissioningPinnacle beam model的每一个特征(feature)都有相应的参数来刻画 Beam model commissioning的过程就是根据测量数据优化这些参数的过程 不同的测量数据，或者说测量数据的不同区域和参数间有对应关系 Pinnacle提供auto-modeling tools，也需要手工修改 Iterative process 42
- 43. Commissioning数据测量AAPM Task Group 106 有详细的guideline: phantom and detector选择，phantom设置，测量过程，数据处理，等 1. Percent Depth Dose 2. Cross beam profiles 3. output factors 4. output Multiple depths @ Field size 1x1 ~ 40x40 cm2 43
- 44. Pinnacle beam model commissioningFocal sourceExtra-focal sourceMLC/Jaw transmission44
- 45. Pinnacle beam model commissioningEnergy spectrumPhoton spectrum generally determines the shape of the depth doseCourtesy of William Song45
- 46. Pinnacle beam model commissioningBeam intensity variation due to Flattening filter (cone shaped)Beam spectrum (软化) variation due to Flattening Filter46
- 47. Pinnacle beam model commissioning电子剂量 （dose due to contamination electrons） Complicated model; less intuituve; Affecting dose up to dmax47
- 48. Pinnacle beam model commissioningMLC Tongue and groove width Rounded leaf tip radius Leaf offset table Inter-leaf leakage etc 48
- 49. Pinnacle beam model commissioningBeam model commissioning的过程是个反复的过程，不断地调整beam model parameters，直到计算的PDD & profiles和测量的数据匹配49红-测量数据 黄-TPS计算
- 50. Life cycle of Linac & TPS新买的Linac Acceptance test – 其目的是确认厂家安装的Linac满足我们预购的指标 (beam energy, beam flatness symmetry, etc) 测很少一组数据 (PDD, profiles)，验证其参数 TPS commissioning – 其目的是通过大量的测量数据，优化TPS的beam model参数，目标是TPS计算的剂量(PDD, profiles)与测量值匹配 一旦确定，基本上不会改变 Routine Linac 质量保证 (Daily, Monthly, Annual) 其目的是检测Linac的输出稳定性；以commissioning data作为参照-这样TPS 计算和Linac输出总能匹配 50
- 51. Data measurementAAPM TG 106 as guideline PDD and beam profiles (两个方向） Field sizes- 2x2, 3x3, 5x5, 10x10, 20x20, 30x30,40x40 cm2 Depth- Dmax, 5, 10, 20, 30 cm Additional PDD 4x4, 6x6, 8x8, 12x12, 15x15, 18x18, 25x25 cm2 Asymmetric MLC beam profile Open Jaw wide to 20 cm Move MLC bank to 0, ±5 cm, ±10 cm, ±15 cm Output factors/TG-51… 51
- 52. 问题?52
该用户的其他文档
相关PPT