ID 109 – A data-driven method for TMS pulse artefact reduction from EEG data: ICA-D

• Published in: Clinical neurophysiology Vol. 127; no. 3; p. e119
• Main Authors: Tomasevic, L., Giambattistelli, F., Siebner, H.R., Rossini, P.M., Tecchio, F.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.03.2016
• Subjects: Neurology
• ISSN: 1388-2457; 1872-8952
• DOI: 10.1016/j.clinph.2015.11.403

Electroencephalographic recording during transcranial magnetic stimulation (EEG/TMS) is used to investigate the state of cortical activity and connectivity (EEG) relative to the non-invasive stimulation of the cortex (TMS). Two technologies have been used so far to acquire the EEG signal during TMS stimulation: the sample-and-hold amplifiers and DC amplifiers with high dynamics. The first bypasses the artefact generated by the pulse itself disabling the acquisition during a short window around the pulse, but with loss of information in that interval. The second records the whole dynamic of the pulse, but there are no solutions yet to deal with a so impacting and unwanted signal. Here we present the procedure that overcomes this limitation: ‘the independent component analysis – decomposition (ICA-D)’. ICA was applied at two levels, on the original data and on the epochs, with identification and automatic reduction of the pulse artefacts from EEG registrations in single subject datasets. This pre-processing procedure returns datasets that are approachable as those obtained from sample-and-hold amplifiers but without the intrinsic disadvantages of that technology. ICA-D is an automatic, data-driven and operator independent procedure. In future it could give new possibilities to unveil cortical responses covered by TMS-related artefacts.