An efficient graph-cut segmentation for knee bone osteoarthritis medical images

The segmentation of bones in the knee region is one of the first essential steps to perform further analysis, classification and osteoarthritis imaging biomarkers discovery. In this paper, an efficient graph-cut based segmentation algorithm is proposed. One of the challenges in current graph-cut schemes is properly distinguishing between regions of interest (ROI) and background regions with features very similar to those of the ROI. Since obtaining a very discriminative cost function is not always feasible, many algorithms require user interaction to provide an extensive number of seed points. In this paper, a new approach is proposed which uses efficient content-based features to achieve segmentation without the need for any user interaction. Experimental results on actual knee MR images demonstrate the effectiveness of the proposed scheme with an average accuracy of 95% using the Zijdenbos similarity index.