Automatic detection of EEG artefacts arising from head movements

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Reliable electroencephalography (EEG) artifact detection is essential for accurate analysis of brain activity.
• Ambulatory EEG systems, like REACT, require robust methods to handle artifacts, particularly those from head movements.
• Existing methods may treat individual artifact components separately, potentially reducing detection efficiency.

Purpose of the Study:

• To investigate optimal features for detecting head movement artifacts in ambulatory EEG data.
• To evaluate the efficacy of a generalized movement artifact class for improved detection.
• To compare the performance of various feature types and statistical methods for artifact identification.

Main Methods:

• EEG data from the REACT ambulatory system was analyzed.
• Temporal, frequency, and entropy-based features were extracted from EEG signals.
• Statistical tests including Kolmogorov-Smirnov and Wilcoxon rank-sum were employed.
• Mutual Information Evaluation Function and Linear Discriminant Analysis were utilized for feature evaluation.

Main Results:

• Significant separation was observed between normal EEG signals and head movement artifacts.
• The proposed generalized movement artifact class demonstrated effective detection capabilities.
• Feature evaluation methods confirmed the distinctiveness of movement-related artifacts.

Conclusions:

• Treating head movement artifacts as a single, generalized class is a viable and effective strategy.
• The identified features provide a strong basis for reliable artifact detection in ambulatory EEG.
• This approach enhances the quality and interpretability of EEG data by minimizing movement-related noise.