Segmentation of brain corpus callosum using graph cuts algorithm based on diffusion tensor imaging

Abstract

The segmentation of white matter was got by using K-means algorithm in order to get the accurate segmentation of corpus callosum from diffusion tensor images.Then the graph cuts algorithm was expanded to tensor space by defining similarity function,and the graph structure was constructed with the tensor similarity to the link after the target and the background seed were selected according to priori.The segmentation of corpus callosum was done through the maximum flow method.The influence of border penalty factor and object seeds on the results was analyzed through the segmentation of the diffusion tensor imaging(DTI) dataset.Results show the correctness of graph cuts algorithm for the segmentation of corpus callosum.