Computer-Aided System based-Deep Learning for Physician Support in Malignancy Score Decision in Breast Ultrasound Imaging

Breast cancer is a significant health concern, and early detection is vital for improving patient outcomes. However,accurate and timely diagnosis remains challenging. This paper proposes a comprehensive system utilizing AI segmentation of breast tumors in ultrasound images to support early detection. The system enhances medical professionals’ capabilities by providing precise and automated tumor segmentation, enabling identification of early-stage tumors and facilitating timely intervention. Classification and segmentation models, improve efficiency and accuracy. The system also addresses the doctor shortage, reducing the burden on healthcare resources. Performance evaluation using a large dataset demonstrates the system’s effectiveness compared to traditional approaches, highlighting the importance of AI-based solutions in early breast cancer detection. Index Terms—Deep Learning, DICOM Viewer, Ultrasound imaging, PACS, Unet, Attention mechanism, Efficient Net