Adaptive successive erosion-based cell image segmentation for p53 immunohistochemistry in bladder inverted papilloma.

Cell nuclei segmentation is a critical issue in automatic cell analysis for cancer diagnosis and prognosis. Marker-controlled watershed segmentation algorithm is used the most commonly. In this paper, adaptive successive erosion-based (ASE) marker extraction method for watershed algorithm is presented, with the goal of extracting markers labelling each individual nucleus, including overlapping cell nuclei. Based on the new marker detection method, an integrated cell image segmentation algorithm is developed for p53 immunohistochemistry in bladder inverted papilloma. Experiments were performed on a number of images, and results demonstrate that the algorithm produces more accurate segmentation than other methods.