Automated segmentation of soft tissue in abdominal CT scans

Abstract

3D models are necessary for many medical procedures. Current techniques require manual segmentation which is tedious and results in coarse models. A current challenge would be to develop automated or semi-automated segmentation techniques to replace this manual segmentation.Atlas-based algorithms are commonly used in medical image segmentatio. Atlases increase the robustness of the algorithm, producing more accurate results. This thesis proposes an automated, non-probabilistic segmentation algorithm for the segmentation of the abdomen from CT slices. The algorithm is designed as a multi-stage pipeline. After pre-processing, the algorithm registers the contours of the atlas to the CT image via a global registration stage and two local registration stages. This thesis also introduces a hybrid active contour called Iterative Corresponding Snake which exhibits greater robustness than the original and is more successful in converging to correct edges in the image.The algorithm was tested for convergence, accuracy and robustness with good results.