ResNet-50: The Deep Networks for Automated Breast Cancer Classification using MR Images
Tejaswini Das, Debasish Swapnesh Kumar Nayak, Anindita Kar, Lambodar Jena, Tripti Swarnkar
Abstract
One in every four cancer cases in women is breast cancer, which is the most prevalent malignancy in this group globally. In 2020, breast cancer will account for one in every eight new instances of cancer, according to projections of 2.3 million new cases. Breast cancer identification, both early and accurate, is critical to improve patient outcomes and survival rates. Magnetic resonance imaging (MRI) has developed as a valuable method for diagnosing breast cancer. Deep learning algorithms, notably convolutional neural networks (CNNs), have demonstrated exceptional effectiveness in a variety of medical image processing applications, including breast cancer classification. The ResNet-50 architecture has received a lot of interest in this context because of its excellent performance in image recognition tasks. ResNet-50 is a deep residual network that introduces skip connections and residual learning to solve the difficulty of training very deep neural networks. A total number of 1480 cancer MRI samples are collected from the Kaggle database, having two classes, which include healthy and malignant scans. It is observed that our proposed model with the data pre-processing techniques has achieved a classification accuracy of 92.01%.