Design of Optimal Experimental Parameters for Diffusion-Weighted MRI Fibre-Tracking Protocols

摘要

In the study of brain development and diseases, diffusion-weighted magnetic resonance imaging (DWMRI) can provide the distribution and orientations of neuron bundles inside elementary volumes, as well as estimates of the diffusion characteristics within brain compartments (axons, extra-axonal medium), based on the acquisition of a series of images in q-space. Various models exist to recover these physiologically relevant parameters, and more are being developed. However, the mathematics underlying the inverse problem has not been systematically investigated yet. For instance, existing q-space sampling protocols are independent of the model used to process the images. This work aims at computing and analyzing the sensitivity coefficients for the fitting parameters of several DWMRI postprocessing methods (DTI, QUAQ) in order to help design optimal experimental protocols.