Deep learning for multi-site ms lesions segmentation: two-step intensity standardization and generalized loss function

Francesca Galassi (1) , Solène Tarride (1) , Emmanuel Vallée (2) , Olivier Commowick (1) , Christian Barillot (1)

1 Empenn - Neuroimagerie: méthodes et applications
2 Orange Labs [Lannion]

Mots clés: MRI; Segmentation; MS lesions; CNN

Domaines: Informatique [cs]; Intelligence artificielle [cs.AI]; Traitement des images [eess.IV]; Sciences du Vivant [q-bio]; Santé publique et épidémiologie; Ingénierie biomédicale; Imagerie

Francesca Galassi, Solène Tarride, Emmanuel Vallée, Olivier Commowick, Christian Barillot. Deep learning for multi-site ms lesions segmentation: two-step intensity standardization and generalized loss function. ISBI 2019 - 16th IEEE International Symposium on Biomedical Imaging, Apr 2019, Venice, Italy. pp.1. ⟨hal-02052250⟩