Investigating the Feasibility of Assessing Magnetization Transfer Properties of Distinct White-Matter Connections

Magnetization transfer ratio (MTR) maps can be associated to the myelin content of the tissue: the higher the MTR, the higher the myelin content. However, in white matter regions where multiple fiber populations, i.e., bundles, can cross the same voxel, the MTR value is voxel- rather than bundle-specific. We propose a method that allows for the assessment of bundle-specific MTR by combining a co-encoded diffusion and MT weighted sequence with Convex Optimization Modeling for Microstructure Informed Tractography (COMMIT), a framework allowing for the estimation of bundle-specific tissue properties. Four healthy subjects (HS) were imaged with a T1w sequence and a novel MT-prepared diffusion-weighted (DW) sequence (MTon). An identical DW sequence, without MT-preparation, was also acquired (MToff). T1 images were segmented in 85 grey matter regions with FreeSurfer and registered to the DW data. A probabilistic tractogram was reconstructed from MToff data and the COMMIT model was then fitted to Mtoff and Mton data separately. Two connectomes, for the Mtoff and Mton data, were calculated by grouping streamlines connecting the same region pair. An MTR weighted connectome was subsequently calculated with element-wise operation on the two connectomes (MTR = (MToff − MTon)/MToff), thus allowing to calculate a bundle specific MTR value. The proposed method was compared to tractometry which, for each streamline in a specific bundle, averages the MTR values along the streamlines path. In all the four HS, in some representative bundles that belong to the left motor network, the MTR values estimated with COMMIT are higher for the bundles connecting the left precentral gyri (L-PrCG) with the medulla (which is a heavily myelinated bundle) than those that connect the L-PrCG with the left subcortical nuclei. In contrast, the tractometry approach appears flat. By applying COMMIT to an innovative dual-encoded MT-dMRI weighted sequence, it is possible to measure bundle-specific MTR.