Feasibility of simultaneous EEG-fMRI at 0.55 T: Recording, Denoising, and Functional Mapping

TL;DR

It is demonstrated that combined EEG-fMRI at 0.55T is feasible and represents a promising environment for multimodal neuroimaging, and the gradient and ballistocardiogram artifacts inherent to this environment are characterized.

Abstract

Simultaneous recording of electroencephalography (EEG) and functional MRI (fMRI) can provide a more complete view of brain function by merging high temporal and spatial resolutions. High-field ($\geq$3T) systems are standard, and require technical trade-offs, including artifacts in the EEG signal, reduced compatibility with metallic implants, high acoustic noise, and artifacts around high-susceptibility areas such as the optic nerve and nasal sinus. This proof-of-concept study demonstrates the feasibility of simultaneous EEG-fMRI at 0.55T in a visual task. We characterize the gradient and ballistocardiogram (BCG) artifacts inherent to this environment and observe reduced BCG magnitude consistent with the expected scaling of pulse-related artifacts with static magnetic field strength. This reduction shows promise for facilitating effective denoising while preserving the alpha rhythm and signal integrity. Furthermore, we tested a multimodal integration pipeline and demonstrated that the EEG power envelope corresponds with the hemodynamic BOLD response, supporting the potential to measure neurovascular coupling in this environment. We demonstrate that combined EEG-fMRI at 0.55T is feasible and represents a promising environment for multimodal neuroimaging.

Figures (6)

• Figure 1: (A) Schematic of the simultaneous EEG-fMRI setup showing synchronization between the MRI trigger and the BrainVision recorder. (B) Protocol overview for Outside, Scanner OFF, and Scanner ON conditions.
• Figure 2: Representative example of EEG signal showing the raw signal across conditions: (A) Outside scanner, (B) Scanner OFF, (C) Scanner ON, and (D) Final cleaned data after artifact subtraction.
• Figure 3: Visual activation and statistical reliability at 0.55T. Representative whole-brain axial montage of t-statistic maps from a single participant, generated from 10 minutes of concatenated task data. Activation maps are overlaid on the 0.55T T1-weighted anatomical image. Significant, contiguous clusters are localized to the primary visual cortex (V1) and surrounding occipital regions. Statistical maps are thresholded at $q < 0.05$ (FDR-corrected) to demonstrate the spatial extent and reliability of the BOLD signal at 0.55T.
• Figure 4: Spectral characterization and task-evoked oscillatory responses at 0.55 T. Power Spectral Density (PSD) analysis across recording conditions (Rest vs. Checkerboard) for occipital channels ($O_z$, $O_1$, and $O_2$) demonstrates the fidelity of neural signal recovery. The spontaneous alpha rhythm (8--13 Hz) is preserved in the resting-state data following artifact subtraction, serving as a baseline indicator of signal integrity. During the visual task, the power spectrum exhibits a robust fundamental response at the 12 Hz stimulation frequency, with distinctly identifiable higher-order harmonics at 24 Hz and 36 Hz. The spectral difference plot (right) highlights the narrow-band power localized to these frequencies, confirming that the denoising pipeline effectively suppresses scanner-induced interference while retaining both physiological and task-specific spectral features.
• Figure 5: Topographical distribution of SSVEP power modulation at 0.55 T. The heatmap illustrates the spatial distribution of the spectral power difference (Checkerboard $-$ Rest) at the 12 Hz stimulation frequency. A highly focal activation is centered over the occipital electrodes ($O_z$, $O_1$, and $O_2$), consistent with the primary generators of the visual evoked response. This localized topography confirms that the signal recovery pipeline suppressed widespread scanner-induced artifacts while preserving the underlying spatial integrity of the neural data in the mid-field environment.