Deep Learning for Improved BreastCancer Detection: ResNet-50 vs VGG16

Breast cancer continues to pose a major global health challenge, underscoring the need for advanced detection and classification methods for mammograms. This study examines the effectiveness of ResNet-50 and VGG16 models in detecting and classifying multiview mammograms. Early and accurate detection of breast cancer is essential for improving patient outcomes and reducing mortality rates. Our approach began with the preparation of mammography images using various image processing techniques, including transfer learning and median filtering. The processed images were then used to train ResNet-50 and VGG16 models for detection and classification tasks. Our experiments demonstrated impressive performance, achieving an accuracy of 96 and an F1 score of 94.66 on the Digital Database for Screening Mammography (DDSM) datasets. These results underscore the potential of deep learning models, particularly ResNet-50, in effectively detecting and classifying multiview mammograms.