Detecting Breast Cancer through Mammography Using Transfer Learning and CNNs

The death toll from breast cancer is still high compared to other female cancers. Improving patient outcomes and maximizing therapy efficacy depends on early and accurate identification. The current study introduces a new method for detecting breast cancer in mammography pictures by combining transfer learning with CNNs. Traditional image analysis methods often struggle with high-dimensional medical images, resulting in misdiagnosis or delayed detection. To address these challenges, pre-trained deep learning models are fine-tuned to extract relevant features from mammograms, enhancing detection accuracy. In order to verify that the suggested strategy works, we look at important performance indicators including sensitivity, specificity, accuracy, and F1-score. The findings show that when compared to models trained from the ground up, transfer learning considerably improves classification performance, with a training time reduction and an accuracy higher than 95%. This approach shows promise in aiding radiologists by providing an automated, reliable diagnostic tool for early detection of breast cancer. Future work will focus on refining the model to handle imbalanced datasets and evaluating its robustness across diverse patient demographics.