Multi-criteria seeded region growing for multi-contrast MRI
ROSSET A, VACHIER C, GARCIA A, VALLEE J
Geneva University Hospital
In many application fields and especially in medical imaging, the segmentation of an object or organ requires to take into account a great number of criteria of several types: morphological, statistical, geometrical... Considering the attributes extraction step solved, the underlying question is that of the adaptation of standard segmentation algorithms acting on gray-scale images to multi-valued images or vectorial images. In this paper, we propose to solve this question with the seeded region-growing algorithm (SRG) available in OsiriX. The procedure we proposed is called multi-criteria seeded region growing (MSRG); it is illustrated and discussed on several MRI examples.