Modeling the Hemodynamic Response Function Using EEG-fMRI Data During Eyes-Open Resting-State Conditions and Motor Task Execution

• DOI: 10.1007/s10548-022-00898-w
• Publication Date: 2022-04-30T08:03:30Z
• Authors: Prokopis C. Proko...; Alba Xifra-Porxas; Michalis Kassinop...; Marie-Hélène Boud...; Georgios D. Mitsis
• Keywords: EEG-fMRI; Blood-oxygen-level dependent

AbstractIn this work, we investigated the regional characteristics of the dynamic interactions between oscillatory sources of ongoing neural activity obtained using electrophysiological recordings and the corresponding changes in the BOLD signal using simultaneous EEG-fMRI measurements acquired during a motor task, as well as under resting conditions. We casted this problem within a system-theoretic framework, where we initially performed distributed EEG source space reconstruction and subsequently employed block-structured linear and non-linear models to predict the BOLD signal from the instantaneous power in narrow frequency bands of the source local field potential (LFP) spectrum (<100 Hz). Our results suggest that the dynamics of the BOLD signal can be sufficiently described as the convolution between a linear combination of the power profile within individual frequency bands with a hemodynamic response function (HRF). During the motor task, BOLD signal variance was mainly explained by the EEG oscillations in the beta band. On the other hand, during resting-state all frequency bands of EEG exhibited significant contributions to BOLD signal variance. Moreover, the contribution of each band was found to be region specific. Our results also revealed considerable variability of the HRF across different brain regions. Specifically, sensory-motor cortices exhibited positive HRF shapes, whereas parietal and occipital cortices exhibited negative HRF shapes under both experimental conditions.