| dc.description.abstract |
Transcranial electrical stimulation (tES) is a non-invasive neuromodulation technique applicable to healthy and diseased subjects that can manipulate brain activity for both therapeutic and research purposes. Simultaneous combination of tES with non-invasive brain imaging techniques might be useful for guiding stimulation parameters to influence brain activity efficiently, and for closed-loop stimulation of the brain. Moreover, such a simultaneous observation is necessary to understand mechanisms underlying tES effects at the network level. However, strong stimulation artifacts at the stimulation frequency make such a simultaneous monitoring by means of MEG or EEG (M/EEG) challenging. At commonly used tES strengths, these artifacts are about 1000 times bigger than brain signals recorded by M/EEG. Therefore, sub-optimal removal of stimulation artifacts leads to residual artifacts that could be mistakenly taken as brain signals. Designing optimal artifact-removal methods requires detailed knowledge about properties of artifacts. In this dissertation, we provide this missing fundamental information by carefully analyzing M/EEG signals during tES. We show that, in contrast to previous assumptions, tES artifacts are non-linearly transformed versions of stimulation currents. This non-linearity manifests itself in both the amplitude and the phase of tES artifacts, and is partly dependent on the stimulation frequency. Specifically, we show that each heartbeat and every respiratory breath strongly modulates both the amplitude and the phase of stimulation artifacts, which makes artifacts dependent on the physiological state of the subject. Due to these modulations, tES artifacts are not narrow band, but contaminate recorded signals even 8 Hz beyond the stimulation frequency. Moreover, the spatial pattern of artifacts continuously varies over time, which decreases the performance of artifact-removal methods based on PCA, ICA or beamforming. In light of our findings, we evaluate available artifact-removal pipelines and show that their outputs are contaminated with residual artifacts, which could have potentially driven biological conclusions made using these pipelines. Finally, we discuss consequences of our findings and provide some ideas for future research regarding how to investigate brain activity during tES. In sum, this dissertation reconsiders assumptions regarding tES artifacts in M/EEG and provides missing fundamental information about their properties. Our results could be used to prevent pitfalls of simultaneous tES and M/EEG and to design and evaluate new artifact-removal pipelines. |
en |
