October 23, 2020

Predicting MEG resting-state functional connectivity using microstructural information

Understanding the role that human brain microstructure plays in the formation of functional connectivity is an important endeavor that could inform therapies for patients with neurological disorders. In this work, magnetic resonance imaging data from ninety healthy participants were used to perform tractography and calculate structural connectivity matrices, using four microstructural metrics to assign connection strength: number of streamlines, fractional anisotropy, radial diffusivity and a myelin measure (derived from the myelin water fraction computed from multi-component relaxometry). Unweighted binarized structural connectivity matrices, marking with 1 existing structural connections and with 0 non-existing ones, and assigning no specific strength to the connections, were also constructed. Magnetoencephalography resting-state data from the same participants were used to calculate their functional connectivity matrices, via correlations of the Hilbert envelopes of the beamformer time series, at four frequency bands: delta (1 – 4 Hz), theta (3 – 8 Hz), alpha (8 – 13 Hz) and beta (13 – 30 Hz). Non-negative matrix factorization (NMF) was then performed to identify the components of the observed resting-state functional connectivity in those bands. Shortest-path-length and search-information analyses of the structural connectomes were used to predict functional connectivity patterns for both the total resting-state functional connectivity and its NMF components. The observed functional connectivity was stronger for brain areas that were closer to each other. The microstructurally-informed algorithms predicted the NMF-derived components of the observed functional connectivity more accurately than they predicted the total (non-factorized) observed functional connectivity. The shortest-path-length algorithm outperformed the search- information algorithm in terms of prediction accuracy. When comparing the weights of the structural connectivity matrices, the number of streamlines and the myelin measure gave the most accurate predictions of functional connectivity, while the fractional anisotropy consistently performed poorly. Overall, different structural metrics paint very different pictures of the human structural connectome, of its relationship to functional connectivity, and of which white matter tracts are implicated in the communication between brain areas.

 bioRxiv Subject Collection: Neuroscience

 Read More

Leave a Reply

%d bloggers like this: