EFFICIENT USE OF CEREBRAL CORTICAL THICKNESS TO CORRECT BRAIN MR SEGMENTATION

Efficient, automatic and robust tools for measurement of cerebral cortical thickness would aid diagnosis and longitudinal studies of neurodegenerative disorders. In this work, we segment a 3D magnetic resonance image of the brain using an expectation-maximization approach. The definition of thickness used is based on the solution of Laplace´s equation in the cortex. Unlike other works, finite difference equations for calculation of cortical thickness are generalized for anisotropic images in order to avoid resampling the input images. We also developed a method which combines information from the thickness estimation with the segmentation probability maps, in order to detect missegmented sulci and correct the segmentation accordingly.