Risks in signal processing pipelines influencing the estimation of phase dependency for EEG-TMS

Phase-dependency of cortico-spinal excitability can be researched using TMS-EEG. Due to the large artifact, non-causal filters can smear the TMS artifact and distort the phase. However, causal filters can become biased by too high filter orders or uneven pass-bands. We explored the influence of different signal processing pipelines on the estimation of the optimal phase. This exploration involved performing two simulation studies. In the first, we simulated two different phase-dependencies (uni-versus bimodal) and sought to recover them with two distinct approaches that have previously been described. In the second, we specifically explored how filter parameters (e.g., order, pass-band) biased the phase estimation. On the basis of these findings, we propose using up-to-date toolboxes, re-running scripts after software updates and performing simulation studies in parallel to safeguard the analysis pipeline of empirical studies.