| Brain activity can be measured non-invasively with functional neuroimaging. EEG/MEG and fMRI in particular complement each other concerning resolution (1 cm2/0.1 ms vs. 1 mm3/0.5 s) and signal sources (electric vs. vascular). Every image pixel represents the activity of about 105 to 107 neurons. Mean field models (MFMs) simplify the description of such neural masses by averaging over neural variability while retaining salient underlying features. However, MFMs that maintain the regional variability and specific connectivity of cortex have so far appeared intractable. We address these impediments through multi-parallel computation and showcase a "proof of principle" forward prediction of co-registered EEG/fMRI for a full-size human cortex in a realistic head model with anatomical connectivity. MFMs usually assume globally uniform neural mass parameters, homogeneous and isotropic long-range connectivity with a single conduction velocity, and simplistic signal expression from cortical source to detector. This allows rapid computation but is unrealistic: different cortical areas vary in their architectonic and dynamical properties, have complex connectivity, and can contribute non-trivially to the measured signal. The approximations are insufficient in particular for spatially resolving brain activity. Our code supports local variation and anatomical connectivity of many thousand triangulation nodes spanning a cortical surface extracted from structural MRI. Proper cortical folding and conduction through a realistic head model is added to obtain accurate signal expression for a proper comparison to experimental data. We predict simultaneously the EEG and BOLD contrast for fMRI by adding an established model for neurovascular coupling and convolving ´Balloon-WindkesselĄ hemodynamics. Furthermore, we incorporate regional connectivity extracted from the CoCoMac database. Importantly, the software will become publically available and can be easily adapted by the user to their preferences. We provide a brief outlook on improving the integration of multi-modal imaging data through fits of a single underlying MFM in this realistic simulation framework, a study we will soon begin with significant experimental support. |
|