Wavelet energy entropy and linear regression classifier for detecting abnormal breasts

Breast abnormalities are the early symptoms of breast cancers. They may also bring in psychoemotional stresses to women. In this study, we developed a new automatic program based on wavelet energy entropy (WEE) and linear regression classifier (LRC): First, we segment region of interest from mammogram images. Second, we calculate WEE from the segmented images. Third, LRC was used as the classifier. We named our method as “WEE + LRC”. The experiment used 10-fold stratified cross validation that was repeated 10 times. The statistical results showed the classification result was the best when the decomposition level was 4, with a sensitivity of 92.00 ± 3.20%, a specificity of 91.70 ± 3.27%, and an accuracy of 91.85 ± 2.21%. The proposed method was superior to other five state-of-the-art methods. In all, our method is effective in detecting abnormal breasts.