Automatic identification of artifacts and unwanted physiologic signals in EEG and EOG during wakefulness. (January 2017)

• Record Type: Journal Article
• Authors: Gerla, V.; Kremen, V.; Covassin, N.; Lhotska, L.; Saifutdinova, E.A.; Bukartyk, J.; Marik, V.; Somers, V.K.
• Is Part Of: Biomedical signal processing and control. Volume 31(2017)
• Page Start: 381
• Page End: 390
• Publication Date: 2017-01
• Subjects: EEG -- Wakefulness -- Artifacts -- Automatic -- Detection -- Identification; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2016.09.006
• Languages: English
• ISSNs: 1746-8094

Abstract

Abstract : Graphical abstract: Abstract : Highlights: Computationally inexpensive methods for robust detection of undesired signals in the EEG. Detection of eye blinks, eye movements, muscle activity and flat line artifacts. Automated, fast and reliable preprocessing of long-term EEG recordings. Verified on real awake EEG data. Abstract: A set of computationally inexpensive methods for reliable and robust detection of undesired signals in the EEG and EOG was designed, implemented, and tested. This strategy includes detection of eye blinking, eye movements, muscle activity, and flat lines in multichannel EEG and EOG data. The proposed methodology was verified on real awake data acquired in controlled conditions (44 recordings of total length 26.38 h) during Maintenance of Wakefulness Tests (MWT). The algorithms worked reliably (average precision was 0.992 ± 0.006, accuracy 0.988 ± 0.006, sensitivity 0.985 ± 0.009, and F1 score 0.988 ± 0.006) and fast (1 h of recording processed in 46.2 ± 5.3 s). We suggest testing this versatile and fast methodology on other type of EEG recordings with modification of threshold parameters if needed. This article reports data from a clinical trials no. NCT01433315 and NCT01580761.