Premium
SU‐E‐J‐85: Anthropomorphic Development for Intermodality Deformation Algorithms Validation
Author(s) -
Vincent S D.,
Archambault L
Publication year - 2012
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4734920
Subject(s) - imaging phantom , computer science , quality assurance , image registration , algorithm , translation (biology) , artificial intelligence , radiation treatment planning , deformation (meteorology) , computer vision , medical imaging , nuclear medicine , radiation therapy , image (mathematics) , medicine , radiology , physics , biochemistry , chemistry , external quality assessment , pathology , messenger rna , meteorology , gene
Purpose: Radiation therapy is often based on a single treatment plan calculated on patient's anatomy at the time of the simulation scan. Deformation algorithms offer the possibility to register initial treatment plan on a daily CBCT. This way, the planning can be adapted to the evolution of patient anatomy. Validation of deformable image registration algorithms (DRA) ideally requires the use of phantoms offering some deformation possibilities Methods: An anthropomorphic, pelvic phantom was built to test volume variation (bladder), deformation of contours (prostate) and translation (all organs). Algorithms must be able to perform intermodality registration. Therefore, images were acquired for both CT and CBCT. The phantom has been created in a way to allow total control of the deformation amplitude. Each of the three types of deformations studied were realized independently and scanned in a manner to have the same initial and deformed images set for each modality. Results: Two algorithm systems were use to compare their efficiency; an open‐source software, a toolbox for registration that offers parameter adjustment and a commercial system with limited control for user. The phantom provides us usable images for DRA validation. For a 2 cm mass center organ translation, the first one reduced 98% of the distance while the other only performed 60%. For a 100 ml volume variation, we get 88% and 62%. Conclusions: Comparison of each intermodal deformation registration performed by the two algorithm systems show how control on parameters improves registration quality. DRA allow the initial planning adaptation on different deformations which occur in human body.