Por João Mendes (IBEB & LASIGE).

Breast cancer, a leading cause of cancer-related morbidity, demands innovative approaches for early detection and risk prediction. The ongoing study focuses on developing a multimodal deep learning (DL) model to predict breast cancer risk using medical images from different modalities such as mammography, digital breast tomosynthesis (DBT), and ultrasound. By integrating these diverse imaging techniques, the project aims to create a system capable of identifying patterns that signal cancer risk, with the potential to personalize screening strategies. The research seeks to classify medical images into "normal," "cancer," and "risk" categories, further stratifying the risk class into sublevels based on disease progression timelines. This stratification allows for more tailored clinical interventions. The project emphasizes the importance of robust data collection, advanced image processing, and cutting-edge DL architectures, all working towards improving early detection accuracy and reducing false positives and negatives. Ultimately, the study aims to contribute to a shift in the clinical paradigm, from reactive treatment to proactive, data-driven prevention.

Short bio: João Mendes is a doctoral candidate in Biomedical Engineering and Biophysics at the Faculdade de Ciências da Universidade de Lisboa (IBEB & LASIGE). His research focuses on applying deep learning techniques to medical imaging, particularly in breast cancer detection and risk prediction.