Nodal analysis of the human brain networks - a new way to couple function and structure

Combining the functional and structural connectomes of the human brain on macro-scale could provide information on various brain diseases and conditions. Unfortunately, accomplishing such combination is not trivial as the connectomes are usually derived from different imaging and measuring modalities that are based on different biophysical effects. The functional connectome is typically obtained using fMRI, EEG,or MEG whereas the structural connectome relies on diffusion-weighted MRI tractography. This work proposes a new way to perform this coupling using an analogy between brain connectomes and electric circuit. The idea is that functional correlations are treated as potentials and structural connections are treated as resistances of a circuit. In a sense, the proposal is a macro-scale version of the Hodkin-Huxley model. This work demonstrates how the proposed function-structure coupling can be applied as filtering for tractography by studying the Default Mode Network of a subject from the Human Connectome Project. To provide a ground-truth example, a phantom simulation is used to detail how the method can solve functional connections in direct and indirect structural connection situations. The method is also applied to a cohort of 207 subjects from HCP to combine their functional and structural connectivity matrices to demonstrate how this approach could be used in group-level analyses. Finally, the proposed model is used to predict subject-wise functional connectomes from structural connectivity data. Predictions matched the measured functional connectomes from the same subjects with approximately Pearson’s r of 0.73 which is on the level or even surpassing many of the state-of-the art graph neural network (GNN) approaches.