Automatic segmentation of head MRI images by knowledge guided thresholding

In this paper we present a procedure which automatically extracts soft-tissue from multi-sliced head MRI images, and then displays three-dimensional shapes of the extracted soft-tissues. We use an iterative thresholding algorithm for segmentation in which an optimum threshold value is decided based on a goodness measure we proposed. This procedure can be used in a preliminary diagnosis in brain surgery without much effort of users, because of whole procedure including threshold selection and segmentation is performed automatically for rendering a three-dimensional image of soft-tissue's surface on a graphic terminal.