Statistical shape model-based segmentation of digital X-ray images

Proposes an improved search procedure for Active Shape Model (ASM) based delineation of anatomical structures in digital X-ray images and compare the original ASM search method with this new technique and a search method optimizing a Bayesian objective function, based on the prior knowledge of the statistical variation of the object boundary points and the variation of image-profiles. The original ASM search method iteratively improves the current estimate of the location of boundary points by a limited least squares adjustment of the pose and shape parameters, the authors' method additionally requires the subsequent changes in shape during the search to be smooth, which is achieved by using a minimum cost path search algorithm, while the Bayesian search methods uses the prior knowledge of the shape-model to constrain the shape parameters within acceptable limits. A number of experiments were performed to compare the accuracy and robustness of all methods using a cross-validation procedure.