Towards the definition of a standard in TMS-EEG data preprocessing.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Biomedical Engineering

Background:

• Combining non-invasive brain stimulation (NIBS) with electrophysiological recordings, like Transcranial Magnetic Stimulation (TMS) and Electroencephalography (EEG), is vital for neuroscience research.
• TMS-EEG integration faces challenges due to magnetic pulse-induced artifacts in EEG data, necessitating robust preprocessing pipelines.
• Existing preprocessing methods for TMS-EEG artifacts show variability, impacting research reproducibility.

Purpose of the Study:

• To quantitatively characterize and compare the performance of three leading TMS-EEG artifact preprocessing pipelines: ARTIST, TESA, and SOUND/SSP-SIR.
• To assess the spatio-temporal precision and reliability of artifact reconstruction for each pipeline.
• To evaluate the impact of each pipeline on inter-trial variability in TMS-EEG data.

Main Methods:

• Utilized a synthetic TMS-EEG signal with a known ground-truth to objectively evaluate artifact removal.
• Applied and compared three distinct preprocessing pipelines: ARTIST, TESA, and SOUND/SSP-SIR.
• Quantitatively assessed pipeline performance based on artifact removal efficiency and signal reconstruction accuracy.

Main Results:

• All evaluated pipelines demonstrated effective artifact removal from TMS-EEG signals.
• Differences were observed in the spatio-temporal precision of reconstructing the original, artifact-free signal.
• The ARTIST pipeline was found to introduce non-intrinsic inter-trial variability into the processed EEG data.

Conclusions:

• While ARTIST, TESA, and SOUND/SSP-SIR are effective for TMS-EEG artifact removal, researchers must consider their differential impact on signal fidelity and variability.
• The choice of preprocessing pipeline can influence the assessment of neural activity and inter-trial consistency.
• This comparative analysis using synthetic data provides a quantitative reference for selecting appropriate TMS-EEG preprocessing methods.