Registration models for tracking organs and tumors in highly deformable anatomies : applications to radiotherapy
(2010)
Files
Thesis_G_Janssens.pdf
Open Access - Adobe PDF
- 23.23 MB
Details
- Authors
- Supervisors
- Macq, Benoit
- Abstract
- (en) In the context of cancer treatment using radiation beams, it is crucial to locate accurately the tumor and organs at risk over time. Therefore, radiotherapy in highly deformable anatomies is made difficult because of tumor and organ motion during treatment. Fortunately, recent imaging techniques, such as respiratory-correlated anatomical and metabolic imaging in 3D, make it possible to observe anatomical deformations. These deformations can then be estimated using non-rigid image registration, the process of transforming different images of the same scene into one common coordinate system. The purpose of this thesis is thus to develop organ and tumor motion estimation methods based on medical image registration, and to integrate them into the radiotherapy treatment process. However, non-rigid registration is a complex and ill-posed problem that requires proper modeling in order to take advantage of prior information, such as manual delineations, bio-mechanical behavior and anatomical constraints, without increasing the time spent by clinicians. In this thesis, we propose and compare several registration models for performing fast, robust and fully automatic motion estimation in the thorax and the abdomen. Several clinical applications that aim at improving radiotherapy treatment by the means of motion estimation are investigated: semi-automatic delineation of moving lung tumors along breathing, deformation of dose distributions, dose planning evaluation taking breathing motion into account and motioncompensation of 4D metabolic images. For each application, we evaluate the accuracy of the proposed registration tools in terms of motion estimation, as well as their ability to fulfill the clinical objectives. Eventually, we present some prospective studies involving other potential use of image registration for improving treatment follow-up and analysis.
- Affiliations
UCLouvain SST/ICTM - Institute for Information and Communication. Technologies, Electronics and Applied Mathematics
Citations
Janssens, G. (2010). Registration models for tracking organs and tumors in highly deformable anatomies : applications to radiotherapy. https://hdl.handle.net/2078.5/131203