Constructing Carbon Fiber Motion-Detection Loops for Simultaneous EEG–fMRI

Abstract

One of the most significant impediments to high-quality EEG recorded in an MRI scanner is subject motion. Availability of motion artifact sensors can substantially improve the quality of the recorded EEG. In the study of epilepsy, it can also dramatically increase the confidence that one has in discriminating true epileptiform activity from artifact. This is due both to the reduction in artifact and the ability to visually inspect the motion sensor signals when reading the EEG, revealing whether or not head motion is present. We have previously described the use of carbon fiber loops for detecting and correcting artifact in EEG acquired simultaneously with MRI. The loops, attached to the subject’s head, are electrically insulated from the scalp. They provide a simple and direct measure of specific artifact that is contaminating the EEG, including both subject motion and residual artifact arising from magnetic field gradients applied during MRI. Our previous implementation was used together with a custom-built EEG–fMRI system that differs substantially from current commercially available EEG–fMRI systems. The present technical note extends this work, describing in more detail how to construct the carbon fiber motion-detection loops, and how to interface them with a commercially available simultaneous EEG–fMRI system. We hope that the information provided may help those wishing to utilize a motion-detection/correction solution to improve the quality of EEG recorded within an MRI scanner.

Towards high-quality simultaneous EEG-fMRI at 7 T: Detection and reduction of EEG artifacts due to head motion

“…In particular, based on the tests conducted with our implementation, a modification with a similar number of sensors and resistor types is expected to have a negligible impact on either EEG or fMRI data quality. As a benefit, a permanent modification would not only save preparation time but also allow for a more geometrically optimal placement of the sensors (Abbott et al, 2014), which here was limited to the positions of the existing EEG electrodes. It should be noted that these sensors are connected to the reference electrode, and are thus not electrically isolated from the scalp.…”

“…Nevertheless, occipital channels still exhibited crucial improvements at a single-trial level, setting a highly encouraging quality mark for this multimodal approach at ultra-high field. Further improvements can be contemplated, for instance through the design of more unified correction schemes that can incorporate gradient, pulse and motion artifact estimation simultaneously -since head motion changes the projection of EEG loops relative to B 0 , it is thereby highly interconnected with the temporal variability of gradient and pulse artifacts, and the three could potentially be more effectively corrected together (Abbott et al, 2014).…”

“…help determining whether particular fluctuations can effectively be attributed to neuronal activity (Abbott et al, 2014). Here, the sensor timecourses were used to reduce motion contributions throughout the EEG timecourses.…”

“…In a later study at 3 T, Masterton et al (2007) proposed the use of multiple sensors based on carbon wire loops, distributed over the EEG cap and sensitive to magnetic induction effects. These sensors were shown to provide richer information for motion and pulse artifact estimation, with clear benefits for the identification of epileptiform activity (Abbott et al, 2014). More recently, socalled reference layer methods have been explored, where EEG-like electrodes are distributed on a conductive layer on top of the EEG cap, but isolated from the scalp, recording only induction effects.…”

Spurious correlations in simultaneous EEG-fMRI driven by in-scanner movement

“…An interesting solution to control for motion on EEG-fMRI data is to record motion co-registered to EEG recordings with a sufficiently high sampling rate. Recently, several motion-recording approaches have been proposed (Abbott et al, 2014;Chowdhury et al, 2014;Jorge et al, 2015;LeVan et al, 2013;van der Meer et al, 2015).…”