abstract: A new approach for quantifying the degree of connectivity between human brain regions from Diffusion Tensor Magnetic Resonance Imaging (DT-MRI) data is presented. To this aim, a functional is proposed and some of its theoretical properties are shown. The connectivity between pairs of white matter points is quantified by minimizing the weighted length of the curves within white matter connecting the points to each other. The weighting factor is a decreasing function of the diffusion coefficient along the curve tangent. This coefficient is a linear function of the diffusion tensor components, which are estimated from DT-MRI data. As a by-product of the analysis, the minimizing curves connecting the two points are provided. The solution of the minimization problem is obtained numerically by approximating the functional on a lattice and then solving a shortest path problem on an undirected weighted graph. The presented method is global and therefore not affected by problems due to fiber branching and crossing. It is also automatic and fast. Some results obtained from the implementation of this method on real data in physiological and simulated pathological conditions are illustrated.