Removing deep brain stimulation artifacts from the electroencephalogram: Issues, recommendations and an open-source toolbox

• Published in: Clinical neurophysiology Vol. 129; no. 10; pp. 2170 - 2185
• Main Authors: Lio, Guillaume, Thobois, Stéphane, Ballanger, Bénédicte, Lau, Brian, Boulinguez, Philippe
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 01.10.2018; Elsevier
• Subjects: Antialiasing; Deep brain stimulation; EEG; Hampel; ICA; Life Sciences; Low-pass filtering; Matched filters; MEG; Neurons and Cognition; Oversampling; Template subtraction; Artifacts; Antialiasing; Matched filters; Hampel; Deep brain stimulation; EEG; ICA; Low-pass filtering; Oversampling; MEG; Template subtraction
• ISSN: 1388-2457; 1872-8952; 1872-8952
• DOI: 10.1016/j.clinph.2018.07.023

•Antialiasing analog filters are not designed to protect against DBS artifacts in EEG.•Low-pass temporal filtering is insufficient: DBS also causes low-frequency artifacts.•Tracking outliers in the frequency domain is necessary to remove DBS artifacts. A major question for deep brain stimulation (DBS) research is understanding how DBS of one target area modulates activity in different parts of the brain. EEG gives privileged access to brain dynamics, but its use with implanted patients is limited since DBS adds significant high-amplitude electrical artifacts that can completely obscure neural activity measured using EEG. Here, we systematically review and discuss the methods available for removing DBS artifacts. These include simple techniques such as oversampling, antialiasing analog filtering and digital low-pass filtering, which are necessary but typically not sufficient to fully remove DBS artifacts when each is used in isolation. We also cover more advanced methods, including techniques tracking outliers in the frequency-domain, which can be effective, but are rarely used. The reason for that is twofold: First, it requires advanced skills in signal processing since no user friendly tool for removing DBS artifacts is currently available. Second, it involves fine-tuning to avoid over-aggressive filtering. We highlight an open-source toolbox incorporating most artifact removal methods, allowing users to combine different strategies.