Automated Detection of Acute Ischemic Stroke on Diffusion MRI in Patients With Suspected Stroke and Stroke Mimics: A Real-World Deep Learning Study - 19/05/26
Doi : 10.1016/j.neuri.2026.100279
PINCET Camille-Albane 1, 2 
, Sandrine DELTOUR 3, 4 , Raphaël Baudrillart 5 , Aurane JEAN-ELIE 1 , Simon ELBAZE 1 , BRUNET D.E.COURSSOU Jean-Baptiste 3, 4 , Milica Mastilovic 6, 7 , Zehira HADDAD 5 , Myriam EDJLALI 8
, Sandrine DELTOUR 3, 4 , Raphaël Baudrillart 5 , Aurane JEAN-ELIE 1 , Simon ELBAZE 1 , BRUNET D.E.COURSSOU Jean-Baptiste 3, 4 , Milica Mastilovic 6, 7 , Zehira HADDAD 5 , Myriam EDJLALI 8 1 Radiology Department, AP-HP, Raymond Poincaré Hospital, 92380, Garches, France
2 Sorbonne Université
3 Neurovascular Department,AP-HP, Raymond Poincaré Hospital, 92380, Garches, France
4 Université de Versailles Saint-Quentin-en-Yvelines, France
5 EPF Engineering School, 55 avenue du Président Wilson 94230 Cachan, France
6 Laboratoire d’imagerie Biomédicale Multimodale (BioMaps), Université Paris-Saclay, CEA, CNRS, Inserm, Service Hospitalier Frédéric Joliot, Orsay, France
7 Faculty of Medicine, University of Novi Sad, Novi Sad, Serbia
8 Multimodal Biomedical Imaging Laboratory (BioMaps), Paris-Saclay University, CEA, CNRS, Inserm, Frédéric Joliot Hospital Department, Orsay, France
Bienvenue sur EM-consulte, la référence des professionnels de santé.
Article gratuit.
Connectez-vous pour en bénéficier!
Sous presse. Manuscrit accepté.
Disponible en ligne depuis le Tuesday 19 May 2026
HIGHLIGHTS |
• | AI triage model trained on a real-world, unselected stroke population |
• | High recall (95%) enabling safe prioritization of urgent cases |
• | Patient-level classification aligned with clinical workflow needs |
• | Demonstrates feasibility of clinically applicable AI beyond simplified datasets |
• | Provides a framework for future multimodal and prognostic AI models |
Plan
© 2026
Publié par Elsevier Masson SAS.
Bienvenue sur EM-consulte, la référence des professionnels de santé.