Patients with Mild Cognitive Impairment (MCI) or Subjective Cognitive Decline (SCD) may or may not develop Alzheimer's disease (AD) dementia. Yet identifying patients at risk is crucial: delaying the onset of the disease by 5 years could reduce prevalence by 50%. To achieve this, we need affordable biomarkers combined with clinically meaningful assessment tools. Current approaches (cognition, imaging or Tau and Amyloid peptide assays) lack precision or specificity (e.g., age-related memory deficits) and involve invasive and costly procedures, sometimes inaccessible in France (e.g., the AT(N) framework). Recently, quantitative diffusion MRI (dMRI) has identified in-vivo gray matter microstructural changes linked to hyperphosphorylated Tau protein, which are of great diagnostic value. Still, we ignore whether and how these changes are responsible for early memory impairment in AD. The MIMA-P project will combine multi-compartment models of the high-resolution diffusion signal with a cognitive assessment of memory based on recent models of medial temporal lobe function to assess the relevance of a new affordable, rapid and non-invasive early marker of the disease.