Smart detection on abnormal breasts in digital mammography based on contrast-limited adaptive histogram equalization and chaotic adaptive real-coded biogeography-based optimization

In this study, we proposed a smart detection method for abnormal breasts in digital mammography. Firstly, preprocessing was carried out to deaden noises, enhance images, and remove background and pectoral muscles. Secondly, fractional Fourier entropy was employed to extract global features. Thirdly, the Welch's t-test was utilized to select important features. Fourthly, the multi-layer perceptron was used as the classifier. Finally, we proposed a novel chaotic adaptive real-coded biogeography-based optimization to train the classifier. We implemented 10-fold cross-validation for statistical analysis. The experimental results showed our method selected in total 23 distinguishing features, and yielded a sensitivity of 92.54%, a specificity of 92.50%, a precision of 92.50%, and an accuracy of 92.52%. This proposed system performs better than five state-of-the-art methods. It is effective in abnormal breast detection.