Breast Cancer Classification Based on Transfer Learning using Mammogram

Although breast cancer has come a long way over the past few decades, women are still facing a lot of anxiety due to a high false-positive rate. So, there is a strong need to reduce unnecessary and burdensome emotional stress for women. Meanwhile, more precise diagnosis can help reduce the number of false-positives. In this study, we propose a method to identify and classify mammogram images into three classes: normal, benign, and malignant, to reduce the rate of false positives and assist the radiologist in making a correct decision. Our approach involves convolutional neural networks, which are a very advanced and efficient technique when dealing with images in deep learning. To provide an efficient solution, transfer learning makes it possible to obtain high performances on small datasets, which were reached with a pre-trained EfficientNet, a state-of-the-art neural network optimized through fine-tuning hyperparameters and data pre-processing. We also used image augmentation techniques to further improve the efficiency of the solution. With the aforementioned architecture and the LAMISDMDB dataset, we achieved an accuracy of 83% and a precision of 83%