Extracting electrophysiological correlates of functional magnetic resonance imaging data using the canonical polyadic decomposition

• Published in: Human brain mapping Vol. 43; no. 13; pp. 4045 - 4073
• Main Authors: Mann‐Krzisnik, Dylan, Mitsis, Georgios D.
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 01.09.2022
• Subjects: Brain - physiology; Brain Mapping - methods; computational modelling; Coupling; Decomposition; EEG; Electroencephalography; Electroencephalography - methods; Electrophysiological Phenomena; Electrophysiology; Feature extraction; Functional magnetic resonance imaging; Hemodynamic responses; Hemodynamics; Humans; Impulse response; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; motor‐imagery; neurovascular coupling; Neurovascular Coupling - physiology; Noise; Noise measurement; Physiological effects; Response functions; resting‐state; Spatial distribution; tensor decomposition; Tensors; magnetic resonance imaging; electrophysiology; motor-imagery; computational modelling; neurovascular coupling; resting-state; tensor decomposition
• ISSN: 1065-9471; 1097-0193; 1097-0193
• DOI: 10.1002/hbm.25902

The relation between electrophysiology and BOLD‐fMRI requires further elucidation. One approach for studying this relation is to find time‐frequency features from electrophysiology that explain the variance of BOLD time‐series. Convolution of these features with a canonical hemodynamic response function (HRF) is often required to model neurovascular coupling mechanisms and thus account for time shifts between electrophysiological and BOLD‐fMRI data. We propose a framework for extracting the spatial distribution of these time‐frequency features while also estimating more flexible, region‐specific HRFs. The core component of this method is the decomposition of a tensor containing impulse response functions using the Canonical Polyadic Decomposition. The outputs of this decomposition provide insight into the relation between electrophysiology and BOLD‐fMRI and can be used to construct estimates of BOLD time‐series. We demonstrated the performance of this method on simulated data while also examining the effects of simulated measurement noise and physiological confounds. Afterwards, we validated our method on publicly available task‐based and resting‐state EEG‐fMRI data. We adjusted our method to accommodate the multisubject nature of these datasets, enabling the investigation of inter‐subject variability with regards to EEG‐to‐BOLD neurovascular coupling mechanisms. We thus also demonstrate how EEG features for modelling the BOLD signal differ across subjects. We show that EEG features of BOLD‐fMRI dynamics can be obtained using the canonical polyadic decomposition. Hemodynamic response functions are also estimated in the process, which allows one to model BOLD‐fMRI signals using EEG signals. We employ our method on simulated data, as well as publicly available task‐based and resting‐state EEG‐fMRI data.