Stroke incidence is increasing and, consequently, so are the number of motor impaired survivors. Whereas patient tailored rehabilitation strategies are believed to greatly improve recovery, a reliable biomarker for gait outcome prediction that enables such patient tailored rehabilitation is currently missing. Therefore, this project aims to explore whether brain structural connectivity between areas responsible for gait can be used as a biomarker for gait recovery prediction. This will be done by characterizing brain connectivity during the first 6 months after stroke using diffusion magnetic resonance imaging (dMRI) and correlating these findings with gait recovery as measured by a comprehensive gait analysis. Brain connectivity will be assessed considering 12 brain areas and 18 white matter pathways between them. Gait analysis will include data on kinetics, kinematics and muscle activity. All results will be used for a machine learning protocol composed by network-based statistics and deep learning As such, when a correlation can be established, the dMRI assessment of connectivity could serve as a biomarker to guide rehabilitation strategies, early in the course of recovery, so that rehabilitation outcome can be improved.