Skip to main content
. 2019 Jan 9;9:1152. doi: 10.3389/fneur.2018.01152

Figure 1.

Figure 1

Workflow of automated reduction of DBS artifacts. A high-order least-square filter, high-pass at 1 Hz was applied first, followed by detection of the peaks of the stimulator artifacts. Second, a principal-component analysis was performed. A large part of the stimulator artifact is accounted for by the first principal component. This component was then deleted and the EEG reconstructed. Third, after an independent component analysis, all resulting components were averaged on the previously detected peaks for identification of the components, including DBS artifacts. These components were deleted and, again, the EEG was reconstructed. Finally, a high order least-square filter, low-pass at 70 Hz, was applied.