Background: Effective connectivity can be explored using direct electrical stimulations in patients suffering from drug-resistant focal epilepsies and investigated with intracranial electrodes. Responses to brief electrical pulses mimic the physiological propagation of signals and manifest as cortico-cortical evoked potentials (CCEP). The first CCEP component is believed to reflect direct connectivity with the stimulated region but the stimulation artifact, a sharp deflection occurring during a few milliseconds, frequently contaminates it.
New method: In order to recover the characteristics of early CCEP responses, we developed an artifact correction method based on electrical modeling of the electrode-tissue interface. The biophysically motivated artifact templates are then regressed out of the recorded data as in any classical template-matching removal artifact methods.
Results: Our approach is able to make the distinction between the physiological responses time-locked to the stimulation pulses and the non-physiological component. We tested the correction on simulated CCEP data in order to quantify its efficiency for different stimulation and recording parameters. We demonstrated the efficiency of the new correction method on simulations of single trial recordings for early responses contaminated with the stimulation artifact. The results highlight the importance of sampling frequency for an accurate analysis of CCEP. We then applied the approach to experimental data.
Comparison with existing method: The model-based template removal was compared to a correction based on the subtraction of the averaged artifact.
Conclusions: This new correction method of stimulation artifact will enable investigators to better analyze early CCEP components and infer direct effective connectivity in future CCEP studies.
Keywords: Artifact removal; Cortico-cortical evoked potentials; Effective connectivity; Electrical cortical stimulation; Intracranial EEG.
Copyright © 2016 The Authors. Published by Elsevier B.V. All rights reserved.