The advent of magnetic resonance (MR) imaging has brought to the field of neurosciences the stupendous ability of in-vivo study of the human brain's tissular properties. More recent developments in the field of computational anatomy have led to automated approaches of volumetric assessment of the brain in voxel-based morphometry (VBM). VBM has provided significant understanding about physiopathology of brain diseases, including psychiatric and neurodegenerative diseases (NDDs). VBM performs tissue classification using algorithms that rely on contrast between tissues and probabilistic maps, termed tissue priors. These algorithms have provided accurate and satisfying study of the human cortex. However, tissue classification of deep gray matter such as the basal ganglia has been found to be largely unreliable. Conventional T1-weighted MR imaging provides lower contrast for deep gray matter than the cortical gray matter. The main reason for this contrast bias is the higher iron concentration in those structures. Moreover, iron deposits increase in the normal ageing adult and reach pathological concentrations in a wide number of neurological disorders including NDDs. Accurate assessment is thus challenged for subcortical structures in both health and disease.
Recently, quantitative MR imaging (qMRI) has been developed to allow quantitative assessment of tissular microstructure of the brain. Those new sequences, such as magnetization transfer saturation (MT) and effective transverse relaxation rate (R2*) parameter maps, provide better contrast by displaying quantitative surrogates for myelin and iron respectively. MT parameter maps have shown to overcome high iron content sensitivity and to be highly suitable for automated delineation of the basal ganglia. Although MT parameter maps provide sufficient contrast, current tissue priors remain insufficient to provide satisfying tissue classification. In this work, we created robust and accurate tissue priors for deep gray matter.