A neural network for breast cancer detection using fuzzy entropy approach

Proposes a novel texture analysis technique based on fuzzy cooccurrence matrix concept, and uses it to deal with early and accurate breast cancer diagnosis by analyzing the microscope-slide biopsy images. A newly proposed feature extraction algorithm is employed to extract the features from the digitized images, then the features are input to a multilayer back-propagation neural network to classify the images into three risk groups. Finally, a resultful comparison of breast cancer diagnosis between the conventional method and the proposed approach is conducted and the conclusion is reached that the proposed method is much superior to the existing methods. The proposed method may have wide applications in the areas of pattern recognition and image processing.