A simultaneous fMRI-EEG acquisition to minimize the MR gradient artifact on human auditory system
Kevin Wen-Kai Tsai1,2, Hsin-Ju Lee2, Ching-Po Lin2, Li-Wei Ko3, Wen-Jui Kuo2, Toni Auranen4, Simo Särkkä5, and Fa-Hsuan Lin6

1Aim for the Top University Project, National Taiwan Normal University, Taipei, Taiwan, 2Institute of Neuroscience, National Yang-Ming University, Taipei, Taiwan, 3Institute of Bioinformatics and Systems Biology, National Chiao Tung University, Hsinchu, Taiwan, 4Advanced Magnetic Imaging Centre, Low Temperature Laboratory, Aalto University, Espoo, Finland, 5Department of Electrical Engineering and Automation, Aalto University, Espoo, Finland, 6Institute of Biomedical Engineering, National Taiwan University, Taipei, Taiwan

Synopsis

Simultaneous fMRI-EEG acquisition provides a good spatial and temporal resolution from MRI and EEG respective to study the human brain function. However, the EEG signal is impaired due to the strong magnetic gradient switching of concurrent MR imaging. A simultaneous interleaved MR InI-EEG recording strategy is proposed to minimize the distortion of the EEG. Our results suggest that the proposed acquisition strategy can reveal similar BOLD contract activation but preserve better auditory evoked potentials than conventional EPI-EEG acquisition.

PURPOSE

Functional MRI (fMRI) cannot detect primary neuronal activity directly but only secondary hemodynamic responses1. To probe the neuronal basis of fMRI responses, it is possible to use simultaneous EEG and fMRI measurements (for review, see 2). However, it has been shown that measuring EEG and fMRI inside the magnet has many technical challenges, including the contamination of signal when MRI gradient coil is switching2. This gradient artifact (GA) has been commonly addressed by subtracting an artifact template, created from averaging fMRI acquisitions, from the contaminated fMRI measurements3. However, subject’s head motion, the jittering between EEG and fMRI measurements, and insufficient EEG dynamic range can limit the performance of such average artifact subtraction. Here we propose to minimize the fMRI acquisition duration and to maximize the EEG acquisition duration in order to measure BOLD signals and high-quality EEG at the same time. Specifically, inverse imaging (InI), a highly parallel fMRI method capable of completing the whole-head measurement with 5 mm resolution at cortex in 0.1 s4, intermittently measured the BOLD signal once every 2 s, in order to minimize GA in concurrent EEG recordings. Experimental results show that our acquisition strategy allows for fMRI similar to that of EPI and event-related potentials measured outside MRI.

METHODS

InI (coronal projection images) only sampled the first 100 ms in 2-s TR. This left a period of 1.9 s (95% of the duty cycle) without artifacts related to MRI magnetization and spatial encoding. Meanwhile, EEG (31 electrodes with impedance < 20 kΩ, reference electrode = FCz) was recorded continuously by a MRI-compatible system (BrainAmp MR Plus, Brain Products GmbH) with 5 kHz sampling rate. Importantly, EEG was temporally synchronized to InI via a 10 MHz clock in the MR scanner. Auditory stimulus (1000 Hz; 200 ms duration) was delivered randomly between 0.2 and 1.4 s after the onset of each TR. There were 50 trials randomly distributed over a 5-minute scan. For comparison, we also recorded the EEG outside MRI with the same auditory stimuli. Conventional multi-slice EPI was also measured concurrently with EEG inside MRI to evaluate how BOLD signal and auditory evoked potentials (AEP) changed by the measurement environment (Figure 1). The InI analysis began by first reconstructing volumetric images the using minimum-norm estimate4., EPI were pre-processed by a customized stream (https://git.becs.aalto.fi/bml/bramila). Both InI and EPI were further analyzed by General Linear Model using canonical models to estimate hemodynamic response. The significance of BOLD signals (t-statistics) were morphed to inflated hemispheres of a standard template (fsaverage in FreeSurfer). The EEG analysis started by first removing gradient artifacts using a MRI artifact template estimated directly from EEG-InI recording3. Subsequently, epochs of EEG were created by taking 0.2 s and 0.5 s before and after each onset of auditory stimuli, respectively. AEP was calculated by taking average over epochs and low-pass filtering (50 Hz) was also applied to AEP.

RESULTS

Figure 2 shows that significant hemodynamic activity was found at the superior temporal gyrus of both hemispheres by EPI and InI. This suggests that both EPI and InI has similar sensitivity and spatial specificity to detect the BOLD signal elicited by auditory stimuli. Figure 3 shows AEPs at the T7 electrode (close to the left temporal lobe) measured outside MRI, inside MRI with EEG, and inside MRI with InI. AEP’s before and after 50-Hz low-pass filtering were shown. These results suggest that while low-pass filtering on EEG-EPI measurements can generate acceptable AEP, we still found significantly distorted AEP waveforms (before N1 and after P2). On the contrary, EEG-InI can avoid such artifacts without losing information at high frequency and much similar AEP to that measured outside MRI.

DISCUSSION

We proposed a simultaneous EEG-fMRI acquisition method by intermittently measuring the BOLD signal (2 s TR) using ultra-fast fMRI acquisition (InI with 0.1 s sampling time) and concurrent EEG recording. High quality neuronal and hemodynamic response were measured. While we used InI in this study, it is possible to replace it with other fast imaging methods, such as simultaneous-multi-slice CAIPI EPI5 to further trade-off between fMRI spatiotemporal resolution and EEG artifacts. In addition to GA, the other prominent noise in EEG acquired inside MRI is the ballistocardiogram (BCG), which is induced EEG signal due to cardiac pulsation. Luckily, BCG is typically below 15 Hz6. In experiments interested in cognitive functions, which are physiologically more related to neuronal oscillation beta (~ 20 Hz) and gamma (> 40 Hz) bands, BCG can be effectively suppressed by low-pass filtering.

Acknowledgements

We thank Dr. Jen-Ren Duann for helpful discussion.

References

1. Logothetis, N. K. What we can do and what we cannot do with fMRI. Nature 2008;453: 869–78.

2. Ullsperger, M. & Debener, S. Simultaneous EEG and fMRI. Oxford University Press, 2010.

3. Allen, P. J., Josephs, O. & Turner, R. A Method for Removing Imaging Artifact from Continuous EEG Recorded during Functional MRI. Neuroimage 2000;12: 230–239.

4. Lin, F.-H., Tsai, W.-K. K., Chu, Y.-H., et al. Ultrafast inverse imaging techniques for fMRI. Neuroimage 2012;62: 699–705.

5. Setsompop, K. et al. Blipped-controlled aliasing in parallel imaging for simultaneous multislice echo planar imaging with reduced g-factor penalty. Magn. Reson. Med. 2012; 67: 1210–1224.

6. Bonmassar, G. et al. Motion and Ballistocardiogram Artifact Removal for Interleaved Recording of EEG and EPs during MRI. Neuroimage 2002;16: 1127–1141.

Figures

Figure 1. The diagram of the interleaved EEG-InI and the continuous EEG-EPI acquisitions. A trigger was sent from the scanner at the beginning of each TR.

Figure 2. Both EPI and InI reveal similar distributions of the BOLD signal elicited by auditory tones.

Figure 3. The auditory evoked potential (AEP) measured by EEG-InI was similar to that measured outside MRI even without low-pass filtering. AEP measured by EEG-EPI with low-pass filtering still had prominent artifacts before N1 and at P2.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3768