Automated recognition method of concave area by using complex PARCOR coefficient and its application to detection of colonic polyps in abdominal X-ray CT images

摘要

The validity of the pattern recognition by using complex PARCOR coefficient was examined as a new way of detecting colonic polyps. In this paper, the fundamental concept of computing the PARCOR coefficient was described at first and a simulation was tried by using some model images with and without concave area mimicking colonic polyps. There were obvious differences between those images. Furthermore, the current work for detecting polyps was described. After the determination of the centerline of colon based on distance transformation technique, the sectional images were calculated based on the normal vector as the direction toward centerlines. The complex PARCOR methods were applied to detect the polyps. As a result of this works, an initial effectiveness by applying the complex PARCOR coefficient was indicated.