Nuclei Segmentation in Breast Histopathology Images using FCM

Recently, computerized methods have been rapidly evolving in the area of digital pathology, with growing applications related to nuclei detection. Cell nuclei segmentation is a fundamental yet crucial step in pleomorphism detection based on the Nottingham Histopathology Grading (NHG) system. The information of the segmented nuclei such as size and morphology properties can be used to determine the scoring decision of the breast histopathology images in terms of pleomorphism. The main purpose of this project is to segment the cell nuclei using Hematoxylin and Eosin (H&E) stained breast histopathology images. In this paper, noise removal associated with Fuzzy C-Mean (FCM) clustering algorithm is introduced which can extract information about each object and then revised and eliminate those irrelevant regions. The RGB input images would be first pre-processed to normalize the color of the input images. The color normalization stage is essential to facilitate the segmentation stage in the later step. For the segmentation process, FCM clustering techniques are applied to better allocate similar pixels into the same clustering while having significant differences between each cluster. Next, the noise region reduction method is performed to discard those pixels which are not related to the properties of cell nuclei. The proposed method is measured by Performance Matrix and the experimental result shows that it demonstrates more desirable performance than the convention FCM clustering method which has average accuracy of 77.30% (±3.332).