Diffusion Tensor Imaging Segmentation by Watershed Transform on Tensorial Morphological Gradient

While scalar image segmentation has been studied extensively, diffusion tensor imaging (DTI) segmentation is a relatively new and challenging task. Either existent segmentation methods have to be adapted to deal with tensorial information or completely new segmentation methods have to be developed to accomplish this task. Alternatively, what this work proposes is the computation of a tensorial morphological gradient of DTI, and its segmentation by IFT-based watershed transform. The strength of the proposed segmentation method is its simplicity and robustness, consequences of the tensorial morphological gradient computation. It enables the use, not only of well known algorithms and tools from the mathematical morphology, but also of any other segmentation method to segment DTI, since the computation of the tensorial morphological gradient transforms tensorial images in scalar ones. In order to validate the proposed method, synthetic diffusion tensor fields were generated, and Gaussian noise was added to them. A set of real DTI was also used in the method validation. All segmentation results confirmed that the proposed method is capable to segment different diffusion tensor images, including noisy and real ones.