Soft Segmentation of CT Brain Data

Unlike research on brain segmentation of Magnetic Resonance Imaging (MRI) data, research on Computed Tomography (CT) brain segmentation is relatively scarce. Because MRI is better at differentiating soft tissue, it is generally preferred over CT for brain imaging. However, in some circumstances, MRI is contraindicated and alternative scanning methods need to be used. We have begun to explore methods for soft tissue segmentation of CT brain data with a goal of enhancing the utility of CT for brain imaging. In this study, we consider the effectiveness of existing algorithms for segmenting brain tissue in CT images. Three methods (Bayesian classification, Fuzzy c-Means and Expectation Maximization) were used to segment brain and cerebrospinal fluid. While these methods outperformed the commonly used threshold-based segmentation, our results show the need for developing new imaging protocols for optimizing CT imaging to differentiate soft tissue detail and for designing segmentation methods tailored for CT.