Histological Image Segmentation With Fuzzy Clustering Method

Abstract

In this study, a method for nuclei image segmentation in histopathological images is proposed. This method is based on a fuzzy clustering method, which is pre-trained on an supplementary domain with very large labeled images, and coupled with an additional network composed of fully connected layers. In this study, Fuzzy Clustering Mean (FCM) was used for clustering and segmentation and the effective ways for breast cancer nuclei detection were obtained. Wherefore, fuzzy clustering means have been used to detect the centers of breast cancer nuclei, then the extracted centers were compared with the ground truth samples. It is worth mentioning, that this work passes through many experimental stages, of detection and segmentation by applying a combination of more than one effective method.