Fuzzy Relevant Regions Segmentation in Breast Histopathology Images using FCM

According to the International Agency for Research on Cancer (IARC), breast cancer has become the most diagnosed cancer in the world. The analysis of breast histopathology images is important. Segmentation of relevant and irrelevant regions is an important pre-processing for the analysis of breast cancer. In the conventional method, the histopathologists need to use the eyeball rolling method to find the tumor regions. The main objective of this paper is to develop an automation segmentation procedure for the relevant regions, which are referred as tumor regions, and irrelevant regions refer as non-tumor regions. The proposed procedure consists of four main stages: (1) color normalization; (2) color model conversion; (3) relevant regions segmentation using FCM, and; (4) masking processing. The proposed procedure was tested using 31 breast histopathology images. The obtained results show that the average accuracy and precision of the relevant region detection are 86.27% (±8.4129) and 84.53% (±10.6636), respectively.