Study On The Non-rigid Registration Technique Of Medical Images And Its Application In Design Of 4D Heavy-ion Radiotherapy Plans

Heavy-ion beam is recognized as the best radiation for radiation therapy due to its inversed depth dose distribution profile and high relative biological effectiveness.In heavy-ion radiotherapy,intra-fractional respiratory movement can cause changes in position and shape of organs.During the stage of treatment plan design,margin needs to be large enough to ensure dose coverage of the target for most of the treatment delivery,which is determined empirically rather than individually.Because of the different respiratory characteristics of various patients,this population-based margin overestimates or underestimates the respiratory range of a patient.Overestimation may cause normal tissues and organs to be involved in the high dose area,thereby damaging the organs at risk;however,underestimation may lead to inadequate exposure to tumor tissue and thus reduce the control rate of tumor.With the advent of 4D CT,the development of 4D heavy ion radiotherapy will be an effective solution to this problem.4D radiotherapy can be defined as the explicit inclusion of the temporal changes in anatomy during the imaging,planning and delivery of radiotherapy.By tracking and compensating for the movement of a target volume during the course of treatment,the dose delivered to the target volume can be maximized while minimizing the damage to the normal tissues around the target volume.Non-rigid image registration is the key technology for the design of 4D radiation therapy plans.By establishing the anatomy of different breathing phase,automatic contour sketching and dose accumulation can be achieved.In this thesis,the non-rigid registration technique of medical images was studied and and its application in design of 4D heavy-ion radiotherapy plan was realized,thereby establishing a substantial basis for developing a 4D heavy-ion treatment planning system.The main research work and results are shown as follows:1)Based on the matlab programming language,the Demons and B-spline based registration algorithms were realized.Using the 4D CT data of 10 patients published online,the accuracies of the two algorithms were compared by evaluating the target registration error of 300 pairs of the corresponding points after registration.The accuracy of the B-spline registration algorithm was found to be superior to that of the Demons registration one.2)The first application of the non-rigid registration technique in 4D radiotherapy planning is contour propagation.In this thesis,the binary-based and triangular mesh-based contour propagation algorithms were realized using the C programming language.The contours of volumes of interest(VOIs)on the CT images at the end-exhale and end-inhale phases for six patients were delineated by a physician.All VOIs were automatically propagated from the end-exhale phase to the end-inhale phase using the two propagation methods.The propagated VOIs were quantitatively compared with the VOIs contoured at the end-inhale phase by the physician.The results showed that the mesh-based algorithm provided superiorities in reliability for contour propagation in 4D radiotherapy.3)The second application of the non-rigid registration techniques is dose reconstruction.Using the matlab programming language,the interpolation-based dose reconstruction algorithm was realized.By comparing dose differences in the cumulative plan calculated by the interpolation-based dose reconstruction algorithm and reference plan,largest difference occurring in the superior-inferior and anterior-posterior direction was found.4)A 4D heavy ion two-dimensional layer-stacking conformal treatment plan was designed and analyzed.Then,two 4D treatment plans targeted as internal target volume(ITV)and organ expanded from gross target volume(GTV)were made and compared.The results showed that the ITV plan did better in protecting normal organizations.