An Integration Scheme for Biomedical CT Image Segmentation

Abstract

The 3D reconstruction of bones is very useful in medical applications such as presurgical planning of prosThesis implant. A vital step in the geometrical modeling of the bone/implant interface is the bone segmentation, which, usually needs a lot of human interaction. In order to release the surgeon from this tedious work, a robust segmentation system is needed. As noted by some authors, the integration of information from different segmentations could give robustness to this process. Here, an automatic integration method for the problem of delineating the contours of bone structures is presented. The system will integrate the classification information obtained from two segmentation methods, one based on region growing and the other based on snakes, using a Markov random field model.