Fuzzy clustering in digital mammograms using Gray Level co-occurrence matrices

Digital mammograms are difficult images to interpret. Data clustering plays a very crucial role in automatic detection of clustered calcifications in digital mammograms. The aim of this paper is to review and compare the performance of the three main data clustering techniques namely K-means clustering, Fuzzy C-Means clustering and Subtractive clustering. The digital mammograms for the study are taken from Mammographie Image Analysis Society (MIAS) digital mammogram database. The contrast limited adaptive histogram equalization (CLAHE) method is used to reduce noise in digital mammograms. The Gray Level co-occurrence Matrices (GLCM) for different distances and angles are constructed. The performance results of the clustering techniques based on mean square errors are tabulated and compared. It was found that the Subtractive clustering technique outperforms the other two techniques.