LASIGE PhD students Hugo Figueiras and José Domingues, and integrated member Nuno Garcia, together with Nuno Matela (IBEB/Ciências ULisboa), published a study in the journal Computers in Biology and Medicine. In the article titled ‘Self-supervised learning for breast cancer detection: A review’, published in the top 10% journal, the authors present a comprehensive analysis of how self-supervised learning (SSL) is transforming breast cancer imaging. The study systematically reviews 50 state-of-the-art works across the full breast cancer detection pipeline, from screening and diagnosis to grading and staging, covering major imaging modalities such as mammography, ultrasound, MRI, digital breast tomosynthesis, and histopathology.

By highlighting how SSL reduces dependence on costly expert annotations while improving robustness and generalisation, the review identifies both current successes and critical research gaps, notably the lack of SSL approaches in breast PET imaging. These findings position self-supervised learning as a key enabler for scalable, clinically relevant AI tools in breast cancer detection and prognosis.

The paper is avalialble here.