Accumulated Dosimetric Investigation Based On Deformable Image Registration In Adaptive Radiotherapy For Cervical Cancer

With the rapid development of medical imaging,the types of imaging equipments or methods that can be used for diagnosis in medicine have increased.Clinical diagnosis and treatment of diseases often need to be combined with multi-modal or the same mode of different periods of images for comprehensive judgment and application,which requires the use of image registration.Pelvic external beam radiotherapy(EBRT)combined with brachytherapy has become the very important mode of radiotherapy for cervical cancer.The tumor control rate and toxicity are closely related to the dose-volume parameters of the tumor targets and organs at risk(OARs).The local control rate of cervical cancer can be improved and the occurrence of complications caused by radiotherapy can be reduced with internal and external irradiation dose accurately distributed and serious attention to the dose-volume limits of the OARs.This study track the deformation and displacement of target areas and OARs by deformable image registration,to realize the dynamic accumulation of internal and external irradiation dose and evaluate the total exposure dose of target area and OARsIn this paper,we systematically review the classification of medical image registration,registration algorithm and methods of evaluating registration accuracy.Firstly,we study the optimal conditions of deformable image registration.Different image quality(with or without applicators)and different registration ranges(target area(S group)/organ structure(M group)/body profile(L group)were applied to conduct the image registration.Dice similarity coefficient(DSC)was applied to evaluated the accuracy of different registration methods.The results show that the optimal CT threshold used to automatically segment the metal applicators is 1700-1800 Hounsfield unit(HU).DSC of the tumor targets and OARs in the group of removing applicators was better than or equivalent to that in the preserving group,and the rectum was the same as that in the group of removing applicators.Among different registration range,M group has the best registration accuracy for OARs as well as good registration accuracy for target areas.Secondly,the influence on dose-volume parameters deviation caused by the reconstructing bias of the applicators was studied.This study simulated the reconstructing bias by changing the value of "offset"(he distance from the top of the applicator to the first source position).The deviation of dose-volume parameters was 0.5%～1.5%,0.5%～3.5%and 0.5%～5.5%respectively due to 1 mm,2 mm and 3 mm reconstructing bias of uterine cavity tube and 0.5%～3.0%,1.0%～6.3%and 1.4%-10.3%respectively due to the corresponding reconstructing bias by insertion needles.In conclusion,with standard reconstructing methods combined with strict quality control,the dose deviation caused by reconstructing bias is limited on dose was limited.Finally,the total dose of radiotherapy was calculated by simple accumulation of DVH parameters from each treatment(S-DVH group)or accumulated by deformable image registration(DIR group).Compare the differences between the two groups,and evaluate the relationship between the cumulative dose of rectum and the severity of radiation proctitis.The results show that there were significant differences in accumulated doses between the two groups,but in acceptable range.The accumulated dose parameters(D2CC/D1C/D01CC))of the rectum was positively correlated with the grade of radiation proctitis.And these parameters could be applied to predict the side effect of the rectum.