Transcranial magnetic stimulation (TMS)-evoked potentials (TEPs) allow one to assess cortical excitability and effective connectivity in clinical and basic research. However, obtaining clean TEPs is challenging due to the various TMS-related artifacts that contaminate the electroencephalographic (EEG) signal when the TMS pulse is delivered. Different preprocessing approaches have been employed to remove the artifacts, but the degree of artifact reduction or signal distortion introduced in this phase of analysis is still unknown. Knowing and controlling this potential source of uncertainty will increase the inter-rater reliability of TEPs and improve the comparability between TMS-EEG studies. The goal of this study was to assess the variability in TEP waveforms due to of the use of different preprocessing pipelines. To accomplish this aim, we preprocessed the same TMS-EEG data with four different pipelines and compared the results. The dataset was obtained from 16 subjects in two identical recording sessions, each session consisting of both left dorsolateral prefrontal cortex and left inferior parietal lobule stimulation at 100% of the resting motor threshold. Considerable differences in TEP amplitudes were found between the preprocessing pipelines. Topographies of TEPs from the different pipelines were all highly correlated (ρ>0.8) at latencies over 100 ms. By contrast, waveforms at latencies under 100 ms showed a variable level of correlation, with ρ ranging between 0.2 and 0.9. Moreover, the test-retest reliability of TEPs depended on the preprocessing pipeline. Taken together, these results take us to suggest that the choice of the preprocessing approach has a marked impact on the final TEP, and that caution should be taken when comparing TMS-EEG studies that used different approaches. Finally, we propose strategies to control this source of variability.

中文翻译：

---

伪影去除方法对TMS-EEG信号的影响

经颅磁刺激（TMS）诱发的电位（TEP）使人们能够在临床和基础研究中评估皮质兴奋性和有效的连通性。但是，由于各种TMS相关的伪影会污染干净的TEP，这些伪影会在TMS脉冲传送时污染脑电图（EEG）信号。已经采用了不同的预处理方法来去除伪像，但是在此分析阶段引入的伪像减少或信号失真的程度仍然未知。知道并控制这种潜在的不确定性来源将提高TEP的评分者间可靠性，并改善TMS-EEG研究之间的可比性。这项研究的目的是评估由于使用了不同的预处理管线而导致的TEP波形的变异性。为了达到这个目的，我们使用四个不同的管道对相同的TMS-EEG数据进行了预处理，并比较了结果。该数据集是在两个相同的记录会话中从16位受试者中获得的，每个会话均由左背外侧前额叶皮层和左下顶叶小叶刺激（静息运动阈值的100％）组成。在预处理流水线之间发现TEP幅度存在很大差异。来自不同管道的TEP的地形在100毫秒以上的等待时间都高度相关（ρ> 0.8）。相比之下，延迟在100 ms以下的波形显示出可变的相关水平，ρ在0.2到0.9之间。而且，TEP的重测可靠性取决于预处理流程。在一起 这些结果使我们认为，预处理方法的选择对最终的TEP有显着影响，在比较使用不同方法的TMS-EEG研究时应谨慎行事。最后，我们提出了控制这种可变性来源的策略。