Breast border extraction and pectoral muscle removal in MLO mammogram images

Abstract

In this paper, we propose a new method for breast border extraction, artifact removal and removal of annotations typically found in the background of mammograms. The proposed method uses adaptive local thresholding to create an initial binary mask for an image. This is followed by the use of morphological operations to remove background artifacts. Then an adaptive algorithm is proposed to automatically detect and remove the pectoral muscle depending on the gray-level intensity values. Preliminary results of experiments conducted on the Mini-MIAS database (Mammographic Image Analysis Society, London, U.K.) show that the proposed method achieves a near 100% success rate for breast contour extraction and the proposed method for pectoral muscle removal achieves nearly 89% accuracy. More importantly, the proposed pre-processing techniques improved the mammogram classification results when compared to using previous pre-processing methods.