Diffusion MRI (dMRI) permits to infer the ensemble average propagator (EAP) from a set of diffusion-weighted images acquired from ng gradient directions and nb b-values. In the context of clinical brain imaging, dMRI sequences seldom exceed 10 minutes acquisition, with ng ≤ 30 and only one b-value. The EAP is then inferred from the resulting low angular resolution diffusion (LARD) images by assuming a Gaussian diffusion profile. In research context, higher angular resolution samplings (ng≥60 and nb≥1) have revealed a non-Gaussian diffusion profile in the white matter when fibers cross. To account for that effect, we propose a non-Gaussian parametric modeling of the EAP, the estimation of which can be accurately performed from LARD images obtained in clinical context.