An application of the supervoxel-based Fuzzy C-Means with a GPU support to segmentation of volumetric brain images

In this paper the problem of segmentation of volumetric medical images is considered. The fast and effective segmentation is obtained by applying the proposed approach which combines the idea of supervoxels and the Fuzzy C-Means algorithm. In particular, Fuzzy C-Means is used to cluster supervoxels produced by the fast 3D region growing. Additional acceleration of the method is achieved with the support of graphical processor (GPU). The detailed description of the proposed approach is given. The results of applying the method to volumetric CT and MRI brain images and CT images of various phantoms are presented, analysed and discussed. The issues related to accuracy of the method, memory workload and the running time are also considered.