K-highest Fuzzy Min-Max Network to Classify Histopathological Images

• Published in: 2019 International Conference on Communication and Signal Processing (ICCSP) pp. 0240 - 0244
• Main Authors: A., Santhos Kumar, Kumar, A., Bajaj, V., Singh, G. K.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-04-2019
• Subjects: Biopsy; Breast cancer; Breast tumor; Breast tumors; Feature extraction; Fuzzy Min-Max network with a K-highest criteria (Kh-FMM); Gray level co-occurrence matrix; Histopathology images; Magnifying factor; Neural networks; Testing
• DOI: 10.1109/ICCSP.2019.8698079

Classification of histopathological images provides great support towards breast cancer diagnosis. Hence, in this work, a Fuzzy Min-Max with a K-highest (Kh-FMM) hyperbox expansion criteria based approach is studied and used to classify the magnification independent breast cancer histopathology images. To categorize the dataset, gray level co-occurrence matrix (GLCM) is utilized to extract features from the images, and are then feed into Kh-FMM neural network. In this paper, BreaKHis dataset is used which contains benign and malignant breast tumor microscopic biopsy images at four different magnifications. Experimental results show that the magnification factor of 200× gives higher accuracy rate than other magnifications.