Segmentation of Medical Images with Regional Inhomogeneities

This paper presents a novel deformable model for accurate delineation of regions of interest in medical images that contain regional inhomogeneities. Such images are common in various medical imaging domains including endoscopy and radiology. The proposed model improves the active contour without edges (ACWE) model by excluding sparse regional inhomogeneities from both the foreground and the background of the images to be segmented. The proposed model is tolerant to noise and allows for the delineation of multiple objects. Experiments were performed on both endoscopic and ultrasonic images from different organs. The results show that the proposed model can be effectively utilized for the delineation of abnormal tissue findings, and in presence of regional inhomogeneities it can be more accurate compared with the ACWE model