Designing and comparing cleaning pipelines for TMS-EEG data: A theoretical overview and practical example.

Combining transcranial magnetic stimulation (TMS) with electroencephalography (EEG) is growing in popularity as a method for probing the reactivity and connectivity of neural circuits in basic and clinical research. However, using EEG to measure the neural responses to TMS is challenging due to the unique artifacts introduced by combining the two techniques. In this paper, we overview the artifacts present in TMS-EEG data and the offline cleaning methods used to suppress these unwanted signals. We then describe how open science practices, including the development of open-source toolboxes designed for TMS-EEG analysis (e.g., TESA - the TMS-EEG signal analyser), have improved the availability and reproducibility of TMS-EEG cleaning methods. We provide theoretical and practical considerations for designing TMS-EEG cleaning pipelines and then give an example of how to compare different pipelines using TESA. We show that changing even a single step in a pipeline designed to suppress decay artifacts results in TMS-evoked potentials (TEPs) with small differences in amplitude and spatial topography. The variability in TEPs resulting from the choice of cleaning pipeline has important implications for comparing TMS-EEG findings between research groups which use different online and offline approaches. Finally, we discuss the challenges of validating cleaning pipelines and recommend that researchers compare outcomes from TMS-EEG experiments using multiple pipelines to ensure findings are not related to the choice of cleaning methods. We conclude that the continued improvement, availability, and validation of cleaning pipelines is essential to ensure TMS-EEG reaches its full potential as a method for studying human neurophysiology.

Assessing cerebellar-cortical connectivity using concurrent TMS-EEG: a feasibility study.

Combined single-pulse transcranial magnetic stimulation (TMS) and electroencephalography (EEG) has been used to probe the features of local networks in the cerebral cortex. Here, we investigated whether we can use this approach to explore long-range connections between the cerebellum and cerebral cortex. Ten healthy adults received single-pulse suprathreshold TMS to the cerebellum and an occipital/parietal control site with double-cone and figure-of-eight coils while cerebral activity was recorded. A multisensory electrical control condition was used to simulate the sensation of the double-cone coil at the cerebellar site. Two cleaning pipelines were compared, and the spatiotemporal relationships of the EEG output between conditions were examined at sensor and source levels. Cerebellar stimulation with the double-cone coil resulted in large artifact in the EEG trace. The addition of SOUND filtering to the cleaning pipeline improved the signal such that further analyses could be undertaken. The cortical potentials evoked by the active TMS conditions showed strong relationships with the responses to the multisensory control condition after ∼50 ms. A distinct parietal component at ∼42 ms was found following cerebellar double-cone stimulation. Although evoked potentials differed across all conditions at early latencies, it is unclear as to whether these represented TMS-related network activation of the cerebellarthalamocortical tract, or whether components were dominated by sensory contamination and/or coil-driven artifact. This study highlights the need for caution when interpreting outcomes from cerebellar TMS-EEG studies.NEW & NOTEWORTHY This is the first study to systematically assess the feasibility of obtaining TMS-evoked potentials from cerebellar stimulation with concurrent EEG. An innovative control condition using electrical stimulation was modified to mimic the sensory aspects of cerebellar stimulation with a double-cone coil, and a state-of-the art cleaning pipeline was trialled. The extent of artifact contamination in signals from stimulation of a cerebellar and an occipital/parietal control site using two TMS coil types was highlighted.

The correspondence between EMG and EEG measures of changes in cortical excitability following transcranial magnetic stimulation.

KEY POINTS: TMS is commonly used to study excitatory/inhibitory neurotransmission in cortical circuits. Changes in cortical excitability following TMS are typically measured from hand (using EMG; limited to motor cortex) or scalp (using EEG); however, it is unclear whether these two measures represent the same activity when assessing motor cortex. We found that TMS-EMG and TMS-EEG measures of motor cortex excitability are differentially affected by sensory confounds at different time points, masking any actual relationship between them in the time domain. In the frequency domain, local high-frequency oscillations in EEG recordings were minimally confounded by sensory artefacts and demonstrated strong correlations with EMG measures of cortical excitability across time, regardless of TMS intensity or waveform. Therefore, despite the effects of sensory artefacts, the two measures of motor cortex excitability share a response component, suggesting that they index a similar cortical activity and perhaps the same neuronal population. ABSTRACT: Transcranial magnetic stimulation (TMS) is a powerful tool for investigating cortical circuits. Changes in cortical excitability following TMS are typically assessed by measuring changes in either conditioned motor-evoked potentials (MEPs) following paired-pulse TMS over motor cortex or evoked potentials measured with electroencephalography following single-pulse TMS (TEPs). However, it is unclear whether these two measures of cortical excitability index the same cortical response. Twenty-four healthy participants received local and interhemispheric paired-pulse TMS over motor cortex with eight inter-pulse intervals, sub- and suprathreshold conditioning intensities, and two different pulse waveforms, while MEPs were recorded from a hand muscle. TEPs were also recorded in response to single-pulse TMS using the conditioning pulse alone. The relationships between TEPs and conditioned-MEPs were evaluated using metrics sensitive to both their magnitude at each time point and their overall shape across time. The impacts of undesired sensory potentials resulting from TMS pulse and muscle contractions were also assessed on both measures. Both conditioned-MEPs and TEPs were sensitive to re-afferent somatosensory activity following motor-evoked responses, but over different post-stimulus time points. Moreover, the amplitude of low-frequency oscillations in TEPs was strongly correlated with the sensory potentials, whereas early and local high-frequency responses showed minimal relationships. Accordingly, conditioned-MEPs did not correlate with TEPs in the time domain but showed high shape similarity with the amplitude of high-frequency oscillations in TEPs. Therefore, despite the effects of sensory confounds, the TEP and MEP measures share a response component, suggesting that they index a similar cortical response and perhaps the same neuronal populations.

The effects of NMDA receptor blockade on TMS-evoked EEG potentials from prefrontal and parietal cortex.

Measuring the brain's response to transcranial magnetic stimulation (TMS) with electroencephalography (EEG) offers unique insights into the cortical circuits activated following stimulation, particularly in non-motor regions where less is known about TMS physiology. However, the mechanisms underlying TMS-evoked EEG potentials (TEPs) remain largely unknown. We assessed TEP sensitivity to changes in excitatory neurotransmission mediated by n-methyl-d-aspartate (NMDA) receptors following stimulation of non-motor regions. In fourteen male volunteers, resting EEG and TEPs from prefrontal (PFC) and parietal (PAR) cortex were measured before and after administration of either dextromethorphan (NMDA receptor antagonist) or placebo across two sessions in a double-blinded pseudo-randomised crossover design. At baseline, there were amplitude differences between PFC and PAR TEPs across a wide time range (15-250 ms), however the signals were correlated after ~80 ms, suggesting early peaks reflect site-specific activity, whereas late peaks reflect activity patterns less dependent on the stimulated sites. Early TEP peaks were not reliably altered following dextromethorphan compared to placebo, although findings were less clear for later peaks, and low frequency resting oscillations were reduced in power. Our findings suggest that early TEP peaks (<80 ms) from PFC and PAR reflect stimulation site specific activity that is largely insensitive to changes in NMDA receptor-mediated neurotransmission.

Characterizing and minimizing the contribution of sensory inputs to TMS-evoked potentials.

BACKGROUND: Transcranial magnetic stimulation (TMS) evokes voltage deflections in electroencephalographic (EEG) recordings, known as TMS-evoked potentials (TEPs), which are increasingly used to study brain dynamics. However, the extent to which TEPs reflect activity directly evoked by magnetic rather than sensory stimulation is unclear. OBJECTIVE: To characterize and minimize the contribution of sensory inputs to TEPs. METHODS: Twenty-four healthy participants received TMS over the motor cortex using two different intensities (below and above cortical motor threshold) and waveforms (monophasic, biphasic). TMS was also applied over the shoulder as a multisensory control condition. Common sensory attenuation measures, including coil padding and noise masking, were adopted. We examined spatiotemporal relationships between the EEG responses to the scalp and shoulder stimulations at sensor and source levels. Furthermore, we compared three different filters (independent component analysis, signal-space projection with source informed reconstruction (SSP-SIR) and linear regression) designed to attenuate the impact of sensory inputs on TEPs. RESULTS: The responses to the scalp and shoulder stimulations were correlated in both temporal and spatial domains, especially after ∼60 ms, regardless of the intensity and stimuli waveform. Among the three filters, SSP-SIR showed the best trade-off between removing sensoryrelated signals while preserving data not related to the control condition. CONCLUSIONS: The findings demonstrate that TEPs elicited by motor cortex TMS reflect a combination of transcranially and peripherally evoked brain responses despite adopting sensory attenuation methods during experiments, thereby highlighting the importance of adopting sensory control conditions in TMS-EEG studies. Offline filters may help to isolate the transcranial component of the TEP from its peripheral component, but only if these components express different spatiotemporal patterns. More realistic control conditions may help to improve the characterization and attenuation of sensory inputs to TEPs, especially in early responses.

The minimal number of TMS trials required for the reliable assessment of corticospinal excitability, short interval intracortical inhibition, and intracortical facilitation.

Transcranial magnetic stimulation (TMS)-induced motor evoked potentials (MEPs) are frequently used to assess corticospinal and intercortical activities. Trial to trial variance of the potentials is commonly observed, and averages of multiple MEPs are usually reported. Multiple trials have resources implications and are not compatible with some experimental protocols. This study investigated the minimum number of MEPs required to reliably assess corticospinal excitability (CSE), short interval intracortical inhibition (SICI) and intercortical facilitation (ICF), within and between sessions. Fifteen healthy volunteers received 35 single-pulse TMS for CSE assessments and 35 paired-pulse TMS for SICI and ICF measurements. Intra- and intersession reliability were examined using intra-class correlation coefficient tests, and stability of the measures was assessed using a general equation estimation analysis. Coefficients of variation were used to probe the effects of inter-individual variability on reliability results. All analyses were carried out on cumulative averages. The optimal number of trials to ensure "excellent" intra and inter-session reliability with low inter-individual variability and the highest level of stability was found to be 20 for CSE and 26 for SICI assessments. Although 30 consecutive trials produced highly reliable ICF measures within a session, inter-session reliability was not significant across 35 trials. These findings have significant implications for improving time efficiency of TMS experiments without compromising intra- or intersession reliability.

The effects of transcranial direct current stimulation on short-interval intracortical inhibition and intracortical facilitation: a systematic review and meta-analysis.

Transcranial direct current stimulation (tDCS) is increasingly being used to affect the neurological conditions with deficient intracortical synaptic activities (i.e. Parkinson's disease and epilepsy). In addition, it is suggested that the lasting effects of tDCS on corticospinal excitability (CSE) have intracortical origin. This systematic review and meta-analysis aimed to examine whether tDCS has any effect on intracortical circuits. Eleven electronic databases were searched for the studies investigating intracortical changes induced by anodal (a) and cathodal (c) tDCS, in healthy individuals, using two paired-pulse transcranial magnetic stimulation (TMS) paradigms: short-interval intracortical inhibition (SICI) and intracortical facilitation (ICF). Additionally, motor-evoked potential (MEP) size alterations, assessed by single-pulse TMS, were extracted from these studies to investigate the probable intracortical origin of tDCS effects on CSE. The methodological quality of included studies was examined using Physiotherapy Evidence Database (PEDro) and Downs and Black's (D&B) assessment tools. Thirteen research papers, including 24 experiments, were included in this study scoring good and medium quality in PEDro and D&B scales, respectively. Immediately following anodal tDCS (a-tDCS) applications, we found significant decreases in SICI, but increases in ICF and MEP size. However, ICF and MEP size significantly decreased, and SICI increased immediately following cathodal tDCS (c-tDCS). The results of this systematic review and meta-analysis reveal that a-tDCS changes intracortical activities (SICI and ICF) toward facilitation, whereas c-tDCS alters them toward inhibition. It can also be concluded that increases and decreases in CSE after tDCS application are associated with corresponding changes in intracortical activities. The results suggest that tDCS can be clinically useful to modulate intracortical circuits.

Crossover design in transcranial direct current stimulation studies on motor learning: potential pitfalls and difficulties in interpretation of findings.

Crossover designs are used by a high proportion of studies investigating the effects of transcranial direct current stimulation (tDCS) on motor learning. These designs necessitate attention to aspects of data collection and analysis to take account of design-related confounds including order, carryover, and period effects. In this systematic review, we appraised the method sections of crossover-designed tDCS studies of motor learning and discussed the strategies adopted to address these factors. A systematic search of 10 databases was performed and 19 research papers, including 21 experimental studies, were identified. Potential risks of bias were addressed in all of the studies, however, not in a rigorous and structured manner. In the data collection phase, unclear methods of randomization, various lengths of washout period, and inconsistency in the counteracting period effect can be observed. In the analytical procedures, the stratification by sequence group was often ignored, and data were treated as if it belongs to a simple repeated-measures design. An inappropriate use of crossover design can seriously affect the findings and therefore the conclusions drawn from tDCS studies on motor learning. The results indicate a pressing need for the development of detailed guidelines for this type of studies to benefit from the advantages of a crossover design.