双种群遗传算法优化射野方向与权重的研究
|
摘要
放射治疗是治疗恶性肿瘤的主要手段之一,其目标是最大限度地将放射线集中照射到肿瘤(靶区),而周围的正常组织及器官应受到最小的剂量。三维适形放射治疗(3D-CRT)使剂量分布形状与肿瘤的形状一致的同时,可使周围正常组织受照射剂量大大减少,从而为靶区的增量照射创造了条件,但是3D-CRT仅能实现与射野照射方向垂直平面内二维方向的适形,为此,近年来发展了的调强放射治疗(IMRT)技术,从而实现了三维方向上的高度适形,由于技术的复杂性,IMRT的优势还远远没有发挥出来,目前尚有许多问题需要解决。其中射野方向及射野权重的优化,是调强放疗中的关键技术。
传统的放射治疗是通过一个基于设计者经验的反复试错方法,不但浪费了大量时间而且最后得到的结果只是一个经验值并不是一个最优解。随着IMRT研究的深入和逆向治疗计划的发展,放射治疗中如何自动选择射野参数引起了广泛的关注,射野方向以及射野权重的优化是二个必不可少的方面,为了节省临床上所需的时间,使放射治疗有效果的同时更具有效率,越来越多的科研人员加入到此项课题的研究行列中来。其中主要用到的优化算法有:线性规划法、均方优化法、梯度法、有约束模拟退火法和遗传算法等。由于临床问题的随机性,以及射野方向与射野权重相互偶合的问题,本文选择了具有全局性、随机性及鲁棒性的双种群遗传算法。
双种群遗传算法(Genetic Algorithm With Two Population,GAWTP)是一类新型全局优化随机搜索技术,它通过向自然界学习,借鉴生物的进化机制来求解问题,特别是,它不要求目标函数具有连续性、导数存在、线性条件等的假设,而且其固有的并行性,使其在速度、优化性能等方面具有良好的性能。
本文利用双种群遗传算法的所有特性,从解决射野方向与射野权重优化问题入手,在三维光子笔射束剂量计算模型下,用二维卷积的方法和快速傅立叶变换(FFT)实现了精确的剂量计算;结合双种群遗传算法对射野方向与权重两个参数进行了优化;优化过程中采用了基于经验的约束方法,并使用了基于剂量约束的目标函数来计算个体适应度的大小。在软件中,搭建了靶区勾画、三维等剂量分布、体积直方图等模块。最后,在此软件平台下,针对同一病例在不同的优化目标下选用不同的优化参数进行了比较。结果表明,用双种群算法同时优化射野方向与权重是一个非常有效的方法,它使剂量的分布更加的适形并且更好的保护了关键器官与正常组织,更有效的节省了放疗时间,从而满足了临床上的要求。
在论文的最后,我们对工作进行了总结,并对遗留的问题进行了讨论与展望。
Radiotherapy plays an important role in oncology, the main object of radiotherapy is that collect the radiation to the GTV(gross target volume), and make the normal tissue to get lowest dosage ,the three-dimensional conformal radiotherapy(3D-CRT) which the high-dose volume matches the targets volume can decrease the dosage of normal tissue, and in this condition, we can increase the clinical dosage , but 3D-CRT conforms only two dimensionally to the target projection in the plane perpendicular to the beam orientation, so that , intensity-modulated radiotherapy(IMRT) is the most advanced from of 3D-CRT,it can produce highly three-dimensionally conformal dose distributions to the target, however, IMRT as a new accurate radiotherapy technology, is so complicated that the difficult problems, including the optimization of beam's orientation and beam's weight.
The conventional radiotherapy base on the experience to choose some parameters, it is not only lost many time but also get the incredible result. To save our clinical time and make out work effective, more and more science to study this problem, and the main optimize algorithms can summarize as follow: the programming quadratic programming, gradient programming, constrained simulated annealing and genetic algorithms are the typical ones. We choose the GAWTP (genetic algorithm with two populations) which has parallel, robust and global superiority to optimize the beam orientations and beam weights
GAWTP is a powerful global optimization approach, it can provide an adaptive method of studding form the outside to solve the real problem, especially, it can also calculate the objective function fast, which is discontinuous, no differentiable, or highly nonlinear because its parallel, robust and global superiority.
In this paper, the dose calculation model based on pencil beams was constructed three-dimensional convolution, and under this calculation model, we optimize the parameters in the same patient with different conditions, it can conclude that the GAWTP was an effective algorithm, It is not only protect on key apparatus but also give the GTV a high-dosage, and most important of all, it can decrease the clinical time. So, the algorithm can be satisfied the clinical requirement
传统的放射治疗是通过一个基于设计者经验的反复试错方法,不但浪费了大量时间而且最后得到的结果只是一个经验值并不是一个最优解。随着IMRT研究的深入和逆向治疗计划的发展,放射治疗中如何自动选择射野参数引起了广泛的关注,射野方向以及射野权重的优化是二个必不可少的方面,为了节省临床上所需的时间,使放射治疗有效果的同时更具有效率,越来越多的科研人员加入到此项课题的研究行列中来。其中主要用到的优化算法有:线性规划法、均方优化法、梯度法、有约束模拟退火法和遗传算法等。由于临床问题的随机性,以及射野方向与射野权重相互偶合的问题,本文选择了具有全局性、随机性及鲁棒性的双种群遗传算法。
双种群遗传算法(Genetic Algorithm With Two Population,GAWTP)是一类新型全局优化随机搜索技术,它通过向自然界学习,借鉴生物的进化机制来求解问题,特别是,它不要求目标函数具有连续性、导数存在、线性条件等的假设,而且其固有的并行性,使其在速度、优化性能等方面具有良好的性能。
本文利用双种群遗传算法的所有特性,从解决射野方向与射野权重优化问题入手,在三维光子笔射束剂量计算模型下,用二维卷积的方法和快速傅立叶变换(FFT)实现了精确的剂量计算;结合双种群遗传算法对射野方向与权重两个参数进行了优化;优化过程中采用了基于经验的约束方法,并使用了基于剂量约束的目标函数来计算个体适应度的大小。在软件中,搭建了靶区勾画、三维等剂量分布、体积直方图等模块。最后,在此软件平台下,针对同一病例在不同的优化目标下选用不同的优化参数进行了比较。结果表明,用双种群算法同时优化射野方向与权重是一个非常有效的方法,它使剂量的分布更加的适形并且更好的保护了关键器官与正常组织,更有效的节省了放疗时间,从而满足了临床上的要求。
在论文的最后,我们对工作进行了总结,并对遗留的问题进行了讨论与展望。
Radiotherapy plays an important role in oncology, the main object of radiotherapy is that collect the radiation to the GTV(gross target volume), and make the normal tissue to get lowest dosage ,the three-dimensional conformal radiotherapy(3D-CRT) which the high-dose volume matches the targets volume can decrease the dosage of normal tissue, and in this condition, we can increase the clinical dosage , but 3D-CRT conforms only two dimensionally to the target projection in the plane perpendicular to the beam orientation, so that , intensity-modulated radiotherapy(IMRT) is the most advanced from of 3D-CRT,it can produce highly three-dimensionally conformal dose distributions to the target, however, IMRT as a new accurate radiotherapy technology, is so complicated that the difficult problems, including the optimization of beam's orientation and beam's weight.
The conventional radiotherapy base on the experience to choose some parameters, it is not only lost many time but also get the incredible result. To save our clinical time and make out work effective, more and more science to study this problem, and the main optimize algorithms can summarize as follow: the programming quadratic programming, gradient programming, constrained simulated annealing and genetic algorithms are the typical ones. We choose the GAWTP (genetic algorithm with two populations) which has parallel, robust and global superiority to optimize the beam orientations and beam weights
GAWTP is a powerful global optimization approach, it can provide an adaptive method of studding form the outside to solve the real problem, especially, it can also calculate the objective function fast, which is discontinuous, no differentiable, or highly nonlinear because its parallel, robust and global superiority.
In this paper, the dose calculation model based on pencil beams was constructed three-dimensional convolution, and under this calculation model, we optimize the parameters in the same patient with different conditions, it can conclude that the GAWTP was an effective algorithm, It is not only protect on key apparatus but also give the GTV a high-dosage, and most important of all, it can decrease the clinical time. So, the algorithm can be satisfied the clinical requirement
引文
[1]胡逸民,肿瘤放射物理学.北京:原子能出版社,1999,538.
[2]L.Jones,P.Hoban.A comparison of physically and radiobiologically based optimization for IMRT[J].Med.Phys.2002,29(7):1447-55.
[3]李宝生,于金明,王立英,等.调强放射治疗计划[J],中国肿瘤,2001,10(8):461-63.
[4]Haken RKT,Kesder M.Use of DVHs,TCPs,and NTCPs in Treatment Planning.In S.Sternick ES(ed).The theory and practice of intensity modulated radiation therapy.Chapter H.Advanced Medical Publishing,1997.
[5]Conway J,Robinson M H.CT virtual simulation[J].Br J Radiol,1997,70 Spec No:S106-118.
[6]Powell,M..Restart Procedures for the Conjugate Gradient Method,in:Mathematical Programming,1977,241-254.
[7]Webb S.Optimization of conformal radiotherapy dose distribution by simulated annealing.Phys.Med.Biol.1989,34:1349-1370
[8]Wu QW,Mohan R,Niemierko A,et al.Optimization of intensity-modulated radiotherapy plans based on the equivalent uniform dose[J].Int J Radiat Oncol Biol Phys.,2002,52(1):224-35.
[9]Pugaehev A B,Boyer A L and Xing L.beam orientation optimization in intensity-modulated radiation treatment planning.Med.Phys.2000,27:1238-1245
[10]陈品,刘三阳,基于双种群遗传策略的组播路由算法[J],电子与信息学报,2002,24(12):1-3
[11]WU XG and ZHU YP.A mixed-encoding genetic algorithm with beam constraint for conformal radiotherapy treatment planning.Med.Phys.2000,27(11):2508-2515
[12]WU XG and ZHU YP.selection and determination of beam weights based on genetic algorithms for conformal radiotherapy treatment planning.Phys.Med.Biol.2000,45:2547-2558
[13]Spirou SV,Foumier-Bidoz N,Yang J,Chui CS,Ling CC.,Smoothing intensity-modulated beam profiles to improve the efficiency of delivery.Med Phys.2001.28(10):2105-12.
[14]胡逸民,肿瘤放射物理学.北京:原子能出版社,1999,180
[15]Perez CA,Bauer M,Edelstein S,et al.Impact for tumor control on survival in carcinoma of the lung trea-ted with irradiation.Int J Radiat Oncol Biol Phys.12:539-547.1996
[16]金浩宇,吕庆文,周凌宏.用解卷积方法提取笔射束核的实现[J].中国生物医学工程学报,2004,23(4):382-5.
[17]Boyer AL,Zhu Y,Wang L,et al.Fast Fourier transform convolution calculation of X-ray isodose distribution in homogeneous media.Med Phys 1989:16(2):248-253.
[18]Chui C.S.Mohan R.Extraction of pencil beam kernels by the deconvolution method.Med.Phys.,1988,15(2):138-144.
[19]Gary A.Ezzell.Genetic and geometric optimization of three-dimensional radiation therapy treatment planning.Medical Physics,1996,23(3):293-305
[20]Pugachev A B,Boyer A L and Xing L.beam orientation optimization in intensity-modulated radiation treatment planning.Med.Phys.2000,27:1238-1245
[21]Langer Met al.A generic genetic algorithm for generating beam weights.Med.Phys.1996,23(6):965-971
[22]M.Langer.R.Brown,S.Morrill,R.Lane,and O.Lee,A genetic algorithm for generating beam intensities.Med.Phys.1994,21:878
[22]Martin A Ebert.Viability of the EUD and TCP concepts as reliable dose indicators[J].Phys Med Biol.,2000,45(2):441-8.
[23]Agren A.,Brahme A.,Turesson I.,Optimization of un-complicated control for head and neck tumors.Int.J.Radiat Oncol Biol Phys.,1990,Vol.19,p1077-1085.
[24]Mohan R.,Mageras G.S.,et al.,Clinically relevant optimization of 3D conformal treatments.Med.Phys.,1992,19:933-944.
[25]A.Niemierko,"Reporting and analyzing dose distributions:A concept of equivalent uniform dose,"Med.Phys.1997,24:103-110.
[26]Jones L.C.and Hoban P.W.,Treatment plan comparison using equivalent uniform biologically effective dose(EUBED).Phys.Med.Biol.2000,45:159-170.
[27]Bortfeld T,Stein J,Preise Kr,Clinically relevant intensity modulation optimization using physical criteria,In:Leavitt DD,Starkschall G,12th International Conference on the Use of Computers in Radiation Therapy.Madison,WI:Medical Physics Publishing;1997,1-4.
[28]Yongjie Li,Jonathan Yao,and Dezhong Yao,Automatic beam angle selection in IMRT planning using genetic algorithm,Phys.Med.Biol.2004:49,1915-1932
[29]M.F.Moiler,"A scaled conjugate-gradient algorithm for fast supervised learning,"Neural Networks 1993,6:525-533.
[30] Kawrakow I. VMC++, electron and photon Monte Carlo calculations optimized for radiation treatment planning. In advanced Monte Carlo for radiation physics, particle transport simulation and applications: Proceedings of the Monte Carlo 2000 Meeting Lisbon, Berlin (2001):229-36.
[31] Seltzer SM. Electron-photon Monte Carlo calculations: the ETRAN code[J]. Appl. Radiat. Isot., 1991,42(7): 917-41.
[2]L.Jones,P.Hoban.A comparison of physically and radiobiologically based optimization for IMRT[J].Med.Phys.2002,29(7):1447-55.
[3]李宝生,于金明,王立英,等.调强放射治疗计划[J],中国肿瘤,2001,10(8):461-63.
[4]Haken RKT,Kesder M.Use of DVHs,TCPs,and NTCPs in Treatment Planning.In S.Sternick ES(ed).The theory and practice of intensity modulated radiation therapy.Chapter H.Advanced Medical Publishing,1997.
[5]Conway J,Robinson M H.CT virtual simulation[J].Br J Radiol,1997,70 Spec No:S106-118.
[6]Powell,M..Restart Procedures for the Conjugate Gradient Method,in:Mathematical Programming,1977,241-254.
[7]Webb S.Optimization of conformal radiotherapy dose distribution by simulated annealing.Phys.Med.Biol.1989,34:1349-1370
[8]Wu QW,Mohan R,Niemierko A,et al.Optimization of intensity-modulated radiotherapy plans based on the equivalent uniform dose[J].Int J Radiat Oncol Biol Phys.,2002,52(1):224-35.
[9]Pugaehev A B,Boyer A L and Xing L.beam orientation optimization in intensity-modulated radiation treatment planning.Med.Phys.2000,27:1238-1245
[10]陈品,刘三阳,基于双种群遗传策略的组播路由算法[J],电子与信息学报,2002,24(12):1-3
[11]WU XG and ZHU YP.A mixed-encoding genetic algorithm with beam constraint for conformal radiotherapy treatment planning.Med.Phys.2000,27(11):2508-2515
[12]WU XG and ZHU YP.selection and determination of beam weights based on genetic algorithms for conformal radiotherapy treatment planning.Phys.Med.Biol.2000,45:2547-2558
[13]Spirou SV,Foumier-Bidoz N,Yang J,Chui CS,Ling CC.,Smoothing intensity-modulated beam profiles to improve the efficiency of delivery.Med Phys.2001.28(10):2105-12.
[14]胡逸民,肿瘤放射物理学.北京:原子能出版社,1999,180
[15]Perez CA,Bauer M,Edelstein S,et al.Impact for tumor control on survival in carcinoma of the lung trea-ted with irradiation.Int J Radiat Oncol Biol Phys.12:539-547.1996
[16]金浩宇,吕庆文,周凌宏.用解卷积方法提取笔射束核的实现[J].中国生物医学工程学报,2004,23(4):382-5.
[17]Boyer AL,Zhu Y,Wang L,et al.Fast Fourier transform convolution calculation of X-ray isodose distribution in homogeneous media.Med Phys 1989:16(2):248-253.
[18]Chui C.S.Mohan R.Extraction of pencil beam kernels by the deconvolution method.Med.Phys.,1988,15(2):138-144.
[19]Gary A.Ezzell.Genetic and geometric optimization of three-dimensional radiation therapy treatment planning.Medical Physics,1996,23(3):293-305
[20]Pugachev A B,Boyer A L and Xing L.beam orientation optimization in intensity-modulated radiation treatment planning.Med.Phys.2000,27:1238-1245
[21]Langer Met al.A generic genetic algorithm for generating beam weights.Med.Phys.1996,23(6):965-971
[22]M.Langer.R.Brown,S.Morrill,R.Lane,and O.Lee,A genetic algorithm for generating beam intensities.Med.Phys.1994,21:878
[22]Martin A Ebert.Viability of the EUD and TCP concepts as reliable dose indicators[J].Phys Med Biol.,2000,45(2):441-8.
[23]Agren A.,Brahme A.,Turesson I.,Optimization of un-complicated control for head and neck tumors.Int.J.Radiat Oncol Biol Phys.,1990,Vol.19,p1077-1085.
[24]Mohan R.,Mageras G.S.,et al.,Clinically relevant optimization of 3D conformal treatments.Med.Phys.,1992,19:933-944.
[25]A.Niemierko,"Reporting and analyzing dose distributions:A concept of equivalent uniform dose,"Med.Phys.1997,24:103-110.
[26]Jones L.C.and Hoban P.W.,Treatment plan comparison using equivalent uniform biologically effective dose(EUBED).Phys.Med.Biol.2000,45:159-170.
[27]Bortfeld T,Stein J,Preise Kr,Clinically relevant intensity modulation optimization using physical criteria,In:Leavitt DD,Starkschall G,12th International Conference on the Use of Computers in Radiation Therapy.Madison,WI:Medical Physics Publishing;1997,1-4.
[28]Yongjie Li,Jonathan Yao,and Dezhong Yao,Automatic beam angle selection in IMRT planning using genetic algorithm,Phys.Med.Biol.2004:49,1915-1932
[29]M.F.Moiler,"A scaled conjugate-gradient algorithm for fast supervised learning,"Neural Networks 1993,6:525-533.
[30] Kawrakow I. VMC++, electron and photon Monte Carlo calculations optimized for radiation treatment planning. In advanced Monte Carlo for radiation physics, particle transport simulation and applications: Proceedings of the Monte Carlo 2000 Meeting Lisbon, Berlin (2001):229-36.
[31] Seltzer SM. Electron-photon Monte Carlo calculations: the ETRAN code[J]. Appl. Radiat. Isot., 1991,42(7): 917-41.