A mechanistic whole brain model to capture simultaneous EEG-fMRI data

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Simultaneous electroencephalography-functional magnetic resonance (EEG-fMRI) data acquisition offers rich insights into brain function but faces challenges due to differing spatiotemporal scales.
• Reconstructing accurate functional connectivity (FC) and its dynamics (FCD) from multimodal data remains a significant hurdle.

Purpose of the Study:

• To introduce a novel oscillatory network model capable of simulating and reconstructing simultaneous EEG-fMRI data.
• To address the spatiotemporal scale mismatch inherent in combining EEG and fMRI.
• To improve the accuracy and computational efficiency of multimodal brain data analysis.

Main Methods:

• A novel oscillatory network model representing brain regions with coupled low-frequency (LFO) and high-frequency (HFO) Hopf oscillators.
• A two-stage training process involving a complex-Hebbian rule for frequency/phase learning and modified backpropagation for amplitude approximation.
• In silico structural perturbation studies to assess the impact of anatomical connectivity changes on brain dynamics.

Main Results:

• The model successfully replicates empirical functional connectivity (FC), FC dynamics (FCD), and modularity across disparate spatiotemporal scales.
• Demonstrated correlation between fMRI FC and EEG frequency band FCs, mediated by LFO-HFO coupling strength.
• Quantified the effects of structural perturbations on network dynamics, including FC, FCD, modularity, and integration levels.

Conclusions:

• The developed oscillatory network model provides a significant advancement in reconstructing simultaneous EEG-fMRI data.
• This framework enhances the understanding of resting-state brain functionality and aids in deciphering neurological disorders across diverse spatiotemporal scales.
• The model's ability to handle cross-frequency interactions and structural perturbations offers a powerful tool for multimodal neuroimaging analysis.