Medical image segmentation using analysis of isolable-contour maps

A common challenge for automated segmentation techniques is differentiation between images of close objects that have similar intensities, whose bound aries are often blurred due to partial-volume effects. We propose a novel a pproach to segmentation of two-dimensional images, which addresses this cha llenge. Our method, which we call intrinsic shape for segmentation (ISeg), analyzes isolabel-contour maps to identify coherent regions that correspond to major objects. ISeg generates an isolabel-contour map for an image by m ultilevel thresholding with a fine partition of the intensity range, ISeg d etects object boundaries by comparing the shape of neighboring isolabel con tours from the map. ISeg requires only little effort from users; it does no t require construction of shape models of target objects. In a formal valid ation with computed-tomography angiography data, we showed that ISeg was mo re robust than conventional thresholding, and that ISeg's results were comp arable to results of manual tracing.