Segmentation of Knee Bones for Surgical Planning

Automatic bones segmentation for surgical planning has a considerable diversity of input domains, such as MRIs and CT scans. This heterogeneity is a challenge for crossmodality algorithms that should equally perform independently of the input image type that is fed to them. MI-Seg is the proposed framework which aims to achieve fair image segmentation of multiple modalities using a single conditional model, e.g. a transformerbased neural network, trained with interleaved mixed data.

Contacts : Matteo Bastico – MAT | David Ryckelync – MAT | Etienne Decencière – CMM) | Laurent Corté – MAT | Yannick Tillier – CEMEF

Centre : PERSEE