Each grey matter voxel in a T1-weighted MRI is modeled by a local circuit of neural masses as proposed by Jansen and Rit [2]. We then define two types of connections between them: local connections (WL) estimated by neural morphological properties [3] and (b) distal connections (WD) characterized using diffusion weighted MRI. The entire model is described by a system of integral differential equation:
where Θ is the differential operator accounting for the synaptic-dendritic dynamics, V is the vector of membrane potentials, S is a sigmoid function, t(d) is the time delay due to distal connections, H, T are the synaptic gains and characteristic time constants and I is the external input. With an appropriate neurovascular coupling and Balloon-Windkessel hemodynamic, the simulated ECoG (25 minutes) can be converted into changes of fMRI BOLD.
Results
Simulation results (see Figure 1) match experimental EEG data from [1], which shows biphasic peaks in the power in all investigated frequencies during the loss and return of consciousness. The simulated distribution of fMRI BOLD predicts strong activity in the left temporal lobe and bilateral middle temporal sulci (see Figure 1).