作者: Pierre-Francois D. D'Haese , Valerie Duay , Rui Li , Aloys du Bois d'Aische , Thomas E. Merchant
DOI: 10.1117/12.480392
关键词: Atlas (topology) 、 Set (abstract data type) 、 Artificial intelligence 、 Radiation treatment planning 、 Segmentation 、 Fully automatic 、 Volume (compression) 、 Normal anatomy 、 Computer vision 、 Computer science 、 Automatic segmentation
摘要: Delineation of structures to irradiate (the tumors) as well be spared (e.g., optic nerve, brainstem, or eyes) is required for advanced radiotherapy techniques. Due a lack time and the number patients treated these cannot always segmented accurately which may lead suboptimal plans. A possible solution develop methods identify automatically. This study tests hypothesis that fully automatic, atlas-based segmentation method can used segment most brain needed plans even tough tumors deform normal anatomy substantially. accomplished by registering an atlas with subject volume using combination rigid non-rigid registration algorithms. Segmented in are then mapped corresponding computed transformations. The we propose has been tested on two sets data, i.e., adults children/young adults. For first set contours obtained automatically have compared delineated manually three physicians. other qualitative results presented.