摘要：To accurately segment putamen with fuzzy edges of basal ganglia region in brain Magnetic Resonance (MR) images, a shortest path based interactive segmentation method is proposed in this paper. Considering anatomical features of putamen, a cost function is designed by using weight coefficients in sub-regions and sub-orientations, and the shortest path search is restrained in a region, so that the fuzzy edges of putamen can be identified effectively. The proposed method solves the problem of Live-wire, which is easy to be attracted by the edge of claustrum in detecting edge of putamen. A great deal of experiments demonstrates that proposed method can realize fast segmentation with accuracy and stability for putamen.
