APPLYING DEEP LEARNING FOR BREAST CANCER DETECTION USING RADIOLOGY

Abstract

Breast cancer remains a critical health challenge across the globe ,where timely identification significantly influences treatment success and survival rates. Traditional diagnostics tools such as mammography, ultrasound imaging, and magnetic resonance imaging(MRI)rely heavily on specialist interpretation, which may introduce inconsistencies, particularly in cases involving dense or complex tissue structures. Recent progress in artificial intelligence has enabled more dependable analysis through data driven models. In this study, a deep learning framework is developed to identify breast cancer from medical image using convolutional neural network (CNN)integrated with a pre-trained Resnet-50 backbone. Prior to model processing, images are refined through denoising, normalization ,and dimensional standardization .The Resnet-50 network is used to derive meaningful visual patterns, which are then passed through classification layers to determine whether a tumor is benign or malignant. To make the predictions more interpretable, Grad-CAM is employed to visually indicate influential regions within the images. The System is built using Python with Tensor flow and Keras, and is connected to a wed interface that allows users to upload images and receive automated diagnostics summaries. The observed outcomes indicate strong predictive capability, suggesting that such systems can support clinical workflows by reducing manual effort and aiding early stage detection