In adaptive proton therapy the patient's anatomy is monitored with periodic daily imaging using computed tomography. This reduces the uncertainty of dose distribution associated with the geometric dependence of the position of the Bragg peak on the anatomy of the patient, which enables planning with tighter margins of the therapeutic volumes. One of the essential steps in adaptive therapy is the propagation of contours of organs, delineated by a clinician on the planning image to the daily images. The automation of this process requires the use of image registration, which entails finding the optimal transformation between two images.
The purpose of this work is to evaluate various deformable image registration parameters and methods, choose the most appropriate ones, find out the similarity of their performance and determine whether registrations methods that were executed on a single patient can be used for other patients. The transformations obtained by image registration are to be used to propagate contours from the reference image to the daily images.
|