Automatic application watershed in early detection and classification masses in mammography image using machine learning methods

Abstract

Mammogram images are used by radiologists for the diagnosis of breast cancer. However, the interpretation of these images remains difficult depending on the type of breast, especially those of dense breasts, which are difficult to read, as they may contain abnormal structures similar to normal breast tissue and could lead to a high rate of false positives and false negatives. In this paper, we present an efficient computer-aided diagnostic system for the detection and classification of breast masses. After removing noise and artefacts from the images using 2D median filtering, mathematical morphology and pectoral muscle removal by Hough's algorithm, the resulting image is used for breast mass segmentation using the watershed algorithm. Thus, after the segmentation, the help system extracts several data by the wavelet transform and the co-occurrence matrix (GLCM) to finally lead to a classification in terms of malignant and benign mass via the Support Vector Machine (SVM) classifier. This method was applied on 48 MLO images from the image base (mini-MIAS) and the results obtained from this proposed system is 93,75% in terms of classification rate, 88% in terms of sensitivity and a specificity of 94%.