AI for health

Automatic kidney localisation in CT volumes using bounding boxes
Samarakoon thesis — Random Regression Forests
The case-based reasoning cycle applied to surgical decision-making
REMI — Case-based reasoning
Image-guided interventional cardiology in the cath lab
Cardiology — deep learning segmentation

AI for health did not arrive in my practice as a trend — it emerged as an answer to concrete clinical needs. I came to it through regression forests, two years before the deep learning explosion, with Prasad Samarakoon’s thesis: a formative experience that taught me, from the outset, to question the data as much as the algorithms. The REMI project then opened a different field: no longer processing images, but reasoning from heterogeneous clinical cases to support surgical decision-making — a deliberately interpretable approach, designed with practitioners. Finally, the Cardiology project saw the adoption of deep learning for organ segmentation, with the same requirement: robust methods, controlled data, and a clear path towards clinical transfer.

Projects & publications

Projets

Publications