Architectural Distortion-Based Mammograms Classification Using Adaptive Contrast Enhancement and Residual Network

A mammogram classification method based on adaptive contrast enhancement combined with residual network is proposed to address the problem that the image features of architectural distortion on mammography images are not obvious and easily overlap with breast glandular tissue and background structures. Firstly, the dataset samples are increased by data augmentation to avoid the overfitting problem caused by insufficient samples in deep learning. Secondly, the adaptive contrast enhancement algorithm is used to improve the image contrast and enhance the image defects. Finally, the residual network model is used for training, testing and validation. The experimental results show that the method has good classification effect, with accuracy and sensitivity reaching 0.8704 and 0.8782, respectively.