Ischemic stroke is the leading cause of acquired disability and the third leading cause of death in industrialized countries. Predicting the lesion evolution observed in the acute phase is a clinical challenge. This question is currently being addressed using magnetic resonance imaging (MRI). Two modalities are mainly used : diffusion imaging which shows the extent of cytotoxic edema and dynamic perfusion imaging which shows the hemodynamic state of the tissues. Diffusion MRI is known to be the image with the greatest predictive potential for detecting the area of irreversible infarct. An open question is the contribution of acute perfusion MRI to achieve this prediction. Several machine learning approaches have been tested and turned out t...