Federated Learning Applications for Breast Cancer

Breast cancer stands as the leading cause of mortality among women worldwide, encompassing all types of cancer. It can affect women of all age groups post-puberty in any country, with its incidence generally rising as women age. Early detection of breast cancer, prior to its spreading, enables more effective treatment and markedly improves survival rates. Employing techniques like mammography, ultrasound, and magnetic resonance imaging (MRI) for breast cancer screening can greatly improve a patient's prognosis and overall outlook.In the realm of medical imaging, there is a growing interest in the utilization of Federated Learning (FL). This approach entails training deep learning models using expansive datasets dispersed across various data centers. Crucially, it upholds privacy by eliminating the requirement to transmit sensitive patient data. The paper aims to analyze current solutions in the state-of-the-art, highlighting workflows and key solutions, taking into consideration the datasets used and architectural choices. Finally, common problems encountered in the works and those typical of the medical domain are analyzed, focusing on possible future solutions to solve them.