Enhanced Breast Cancer Detection in Multiple Imaging Modalities using Deep Learning

According to statistics, there are around 2.1 million new cases of breast cancer each year, making it the most common kind of cancer among women worldwide. Approximately 15% of female cancer fatalities in 2018 were believed to be caused by breast cancer, which claimed 627,000 lives. For diagnosis and prompt treatment, it is crucial to have an accurate analysis of medical pictures. Breast cancer has been classified using different methods. Recent years have seen an increase in interest in the use of deep learning algorithms for breast cancer categorization due to their high accuracy in prediction compared to conventional methods. In this study, three different datasets that were screened using ultrasound, mammography, and histopathology procedures were employed. A customized and fully connected CNN with data augmentation is built to have enhanced accuracy. The features were trained after being retrieved. Inception Net transfer learning model was used to further boost accuracy. As a result, the model’s accuracy for all three datasets was raised to more than 94%.