Detection and classification of subject-generated artifacts in EEG signals using autoregressive models

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Manual identification of artifacts in continuous electroencephalogram (EEG) recordings is time-consuming and impractical for large datasets.
• Subject-generated artifacts significantly impact EEG signal analysis.
• Developing automated methods for artifact detection and classification is crucial for reliable EEG studies.

Purpose of the Study:

• To develop and evaluate an automated method for accurate detection and classification of artifacts in continuous EEG recordings.
• To assess the effectiveness of autoregressive (AR) models and support vector machine (SVM) classifiers for EEG artifact analysis.

Main Methods:

• Feature extraction using autoregressive (AR) models to characterize EEG signals with artifacts.
• Utilizing scale-invariant AR model parameters as features for artifact classification.
• Employing a support vector machine (SVM) classifier to discriminate between different artifact conditions.

Main Results:

• The proposed method achieved reliable classification of several different artifact conditions across subjects with approximately 94% accuracy.
• AR model parameters proved effective as features for discriminating artifact signals.
• The approach demonstrated consistency in artifact discrimination both within and across individuals.

Conclusions:

• Autoregressive (AR) modeling combined with support vector machine (SVM) classification offers a robust and accurate solution for automated EEG artifact detection and classification.
• This methodology can significantly improve the efficiency and reliability of analyzing large EEG datasets.
• The findings suggest that AR modeling is a valuable tool for distinguishing artifact signals in EEG recordings.