This report describes and evaluates the steps needed to perform modern model-based interpretation of the corpus callosum in MRI. The process is discussed from the initial landmark-free contours to full-fledged statistical models based on the Active Appearance Models framework. Topics treated include landmark placement, background modelling and multi-resolution analysis. Preliminary quantitative and qualitative validation in a cross-sectional study show that fully automated analysis and segmentation of the corpus callosum are feasible.
|Publication status||Published - 2003|