Automatic Multi-objective Optimization For Intensity Modulated Radiation Therapy

Radiotherapy is one of the three techniques of treating cancer in the world.About 70%of patients need different degrees of radiotherapy at different stages of their disease,so it is of great importance to promoting the development of radiotherapy techniques and improving the treatment of radiotherapy.At present,the most widely used radiotherapy technique in clinic is Intensity Modulated Radiation Therapy(IMRT),and its plan is designed by reverse planning,so it can effectively ensure the coverage and uniformity of plan target volume while sparing the surrounding normal tissue and OARs,it is useful to improve the treatment of radiotherapy.In IMRT,the plan quality is important,and it effects the treatment of radiotherapy.Even for a patient,multiple goals(or dose constraints)often need to be optimized simultaneously in the planning optimization.Meanwhile,there are fundamental conflict between the need to reduce dose to OAR while maintaining target homogeneity.Besides,the ideal values for these dose constraints are unknown prior to planned design.In order to a best treatment,we must find a 'best' trade-off between these dose constraints.In clinic,optimizations are often performed in a try-and-error manner by radiotherapy physician or dosimetrist until an suitable plan is found,the quality of plan depends primarily on the time,effort spent in trial and their own design experience,there are significant differences in plan quality and the planning efficiency is low.This paper systematically analyzes the reasons of the differences in planning quality and the inefficiency of planning.In order to ensure the plan quality and improve the efficiency of planning,the following work has been done:First,an automatic voxel weighting factor-based IMRT multi-objective optimization method is proposed to solve the large difference in plan quality caused by the contradiction between multiple objectives.The method generated a constraint priority list based on the clinical importance of dose constraint at the beginning.Then,it relaxed and tightened dose constraints based on the list until a plan with the property that tightening a single constraint is only possible if at least one other constraint violation is found.In addition,the method uses an automatic voxel weighting factor-based FMO optimization model,compared to organ weighting factor-based method used in clinic,the new model expands the search space of optimal solution,so it is easy to obtain a better plan.The DVH(Dose Volume Histogram)and endpoints results of 10 patients with cervical cancer showed that the automatic voxel weighting factor-based IMRT multi-objective optimization plan are superior to the original(clinical)plan,while the coverage and dose homogeneity of PTV is no worse than the original plan,OARs is exposed to less dose.Second,a patient-specific prediction guided automatic treatment planning optimization method for IMRT is presented to overcome the low efficiency of planning caused by the value of the ideal dose constraint is not known before the planning.It is an improvement of automatic voxel weighting factor-based IMRT multi-objective optimization method.This method predicted the dose constraint value before the planning,and then use the reasonable dose constraint value in optimization.This paper compared the 10 GYN patients result of the patient-specific prediction guided automatic treatment planning optimization and original plan,the patient-specific prediction guided automatic treatment planning optimization and the automatic voxel weighting factor-based IMRT multi-obj ective optimization(including DVH and endpoints),it show that the patient-specific prediction guided automatic treatment planning optimization can ensure the coverage and uniformity of PTV while sparing the OARs.Also,from the 10 GYN patients result of adjusting dose constraints,the patient-specific prediction guided automatic treatment planning optimization decreased the number of adjusting dose constraints,planning efficiency improved.This paper is mainly about the problem of the large difference in plan quality and the low efficiency of planning.To solve the problem,an automatic voxel weighting factor-based IMRT multi-objective optimization algorithm is implemented to ensure the plan quality.And an improved patient-specific prediction guided automatic treatment planning optimization for IMRT is proposed to ensure the plan quality and improve the efficiency of planning.Although some preliminary results have been obtained in this study,further studies are needed for clinical application.