Microstructure-weighted connectomics: validation and application of a novel diffusion MRI protocol for mapping brain structural connectivity

Diffusion Magnetic Resonance Imaging is an advanced MRI technique that maps the three-dimensional diffusion of water molecules within brain tissue. This technique is particularly effective for reconstructing White Matter fibre pathways and brain structural connectivity. In this work, two multi-shell diffusion acquisition protocols were used: the standard extended NODDI protocol, which employs two b-values (“shells”), and the novel connectome protocol, with four b-values. The additional shells in the connectome protocol provide enhanced details of brain tissues, which are crucial for accurate connectome reconstruction. The novel protocol was tested and validated using both phantom and in-vivo data acquired with 3T scanners at two sites. Data were analysed by fitting Diffusion Tensor Imaging (DTI) and Bingham-NODDI models. Phantom results demonstrated excellent temporal stability for each protocol and strong agreement between them, with Coefficients of Variation (CVs) below 2%. In-vivo results confirmed the accordance between the protocols and showed good inter-site reproducibility, with no systematic bias detected in the Bland-Altman analysis. Following these validations, microstructure-weighted connectomes were reconstructed from the connectome protocol data. The connectomes were modelled as 3D networks, with edges weighted by Fractional Anisotropy and Mean Diffusivity indices derived from DTI fit, and by Intra-Neurite and Extra-Cellular Volume Fractions from Bingham-NODDI fit. Network metrics extracted from FA-, MD- and INVF-weighted connectomes showed low CVs and no systematic bias across sites, except for modularity. The ECVF-weighted connectome exhibited slightly worse inter-site reproducibility. In summary, these findings highlight the stability and reliability of the connectome protocol and the robustness of the connectome construction process, validating their use for investigating connectome properties in neurological diseases such as Multiple Sclerosis.