Streamlined Breast Cancer Identification: Self-attention CNN with Momentum Search Optimization

Abnormal discontinuities in the connective tissue cells of the female’s feeding ducts are indicative of breast cancer. When indications of breast cancer appear in the milk ducts, a significant number of women passed away from the dis-
ease. The death rates may drop when the determination is detected earlier. It takes a lot of time for oncologists and radiologists to manually analyze mammogram pictures for breast cancer. To avoid tedious analysis and streamline the classification process, our research work proposed hybrid deep learning based Conv Neural Network with Momentum Search Optimization Approach for classifying tumors and non-tumors in mammogram images. Themammography pictures undergone image preprocessing using seam carving approach comprises phases of masking, cropping, rotating, and flipping. Following the pooling and fattening layer, the characteristics were gathered individually during the initial classification step. Additionally, the characteristics are supplied as input to the fully connected layer of the proposed CNN-MSOA model. Our experimental outcomes demonstrate that hybrid CNN-MSOA model attained 99.13% accuracy using CBIS DDSM dataset. Moreover various metrics were evaluated as mentioned in experimental part which predicts the performance of model in breast cancer diagnosis. We show the benefits of our proposed algorithm over the state-of-the-art approaches, especially in terms of accuracy, Precision, recall, F score and ROC AUC score.