Organ segmentation is a vital task in diagnostic medicine. The ability to perform it automatically can save clinicians time and labor. In this paper, a method to achieve automatic segmentation of organs in three-dimensional (3D), non-annotated, full-body magnetic resonance (MR), and computed tomography (CT) volumes is proposed. According to the method, training volumes are registered to a chosen reference volume and the registration transform obtained is used to create an overlap volume for each annotated organ in the dataset. A 3D probability map, and its centroid, is derived from that. Afterwards, the reference volume is affinely mapped onto any non-annotated volume and the obtained mapping is applied to the centroid and the organ probability maps. Region-growing segmentation on the non-annotated volume may then be started using the warped centroid as the seed point and the warped probability map as an aid to the stopping criterion.
展开▼
