Mammography Image Analysis Using Wavelet and Statistical Features with SVM Classifier

Abstract

Breast Cancer is one of the leading causes for death. Early detection is the only way to prevent the breast cancer. Mammography is basic screening test for breast cancer. It is low level X-ray imaging with less cost and more effectiveness. This paper aims to design an automated analysis system for breast cancer detection and classification. The proposed system works in three stages, pre-processing, segmentation and classification. In preprocessing, thresholding and region growing technique used to remove the background and pectoral muscle respectively then Median filter and Contrast limited adaptive histogram equalization (CLAHE) used to enhancing the quality of the image. Tumor segmented by contour based segmentation technique then support vector machine (SVM) classifier used discriminate the benign from malignant with statistical features extracted from level 4 decomposition of wavelets such as Haar, Daubechies (db4), Coiflet and Bi-orthogonal (bior 2.8). Among these wavelet features the db4 features effectively classify the tumor type with high accuracy, specificity and sensitivity as 96, 97.30, 92.31% respectively. The analysis of proposed method conducted on MIAS dataset and the results are promising.