Computer-aided detection of breast cancer using pseudo zernike moment as texture descriptors

Abstract

The breast cancer is a prominent cause of decease in women worldwide. The early detection of breast cancer may avoid the causing symptoms to spread beyond the breast which can significantly reduce the decease rates. In this paper, we develop a computer-aided diagnosis (CAD) system to detect and classify the abnormalities. The input region of interest (ROI) is manually extracted and subjected to further several preprocessing steps. The pseudo zernike moment (PZM) is used for feature extraction as a texture descriptors. A support vector machine is implemented to classify the extracted features accordingly. The proposed system accomplished overall accuracy of 93.63% with 92.14% sensitivity and 94.14% specificity. The area under the curve (AUC) is found to be 0.974.