Deep Learning for Mammography Classification: A CNN Approach on CBIS-DDSM

It has widely been reported that breast cancer is one of the main causes of death among women throughout the entire world. However, its premature detection coupled with an appropriate treatment at the right time can certainly increase the chances of recovery and survival. The present work aims to propose an approach that is based on a convolutional neural network (CNN) for the automatic classification of mammograms from the open-source mammography database CBIS-DDSM. This CNN model was next trained on small size images (50×50×3) for the purpose of optimizing the computation time while ensuring high degree of accuracy. Then, once the model was trained for 50 epochs, it was able to classify benign and malignant tumors, with a test accuracy of 85.04%.