Abnormal breasts are one of the first signs of breast cancer that can put a psychological burden on women. This is my implementation of the paper providing a novel method to differentiate between normal and abnormal breasts based on Wavelet Energy Entropy (WEE) and linear regression classification (LRC).
The preprocessing diagram:
In brief, the region of interest (ROI) is initially identified from digital mammograms. The ROI's WEE is then calculated, and the images are then classified using the LRC classifier. According to 10-fold cross-validation, the best classification results occur when the decomposition level is 4. Sensitivity is 82%, specificity is 81.5%, and accuracy is 81.23%. (Compared to the original paper, my results are lower.)
Wavelet decomposition:
