Mammogram Classification using Supervising Vector Machine and K-Nearest Neighbors for Diagnosis of Breast Cancer

Abstract

Breast cancer attacks women in their early productive years of life which become a public health problem, but if detected earlier it will be cured out with limited resources, while retching the advanced stage treating disease is too expensive and often poor outcome. The aim of this research is to obtain a method to classify the breast into either normal or abnormal tissues. The proposed method which is produced in this paper, is incorporating the Local Contrast (LC) with the Contrast Limited Adaptive Histogram Equalization (CLAHE), that will increase the contrast enhancement and to improve the appearance of the image. Region growing technique is used to extract and crop the region of interest (ROI), that contains the tumor with the texture features of that region automatically, with the help of using the Gray Level Co-occurrence Matrix (GLCM) technique. These features are fed into the Fine Gaussian Supper Vector Machine (SVM) classifier. As observed from the performance evaluation the proposed method classifies the mammography images with 97 % accuracy, 95% specificity and 98 % sensitivity.