Related Experiment Videos
Context related artefact detection in prolonged EEG recordings
M van de Velde1, I R Ghosh, P J Cluitmans
1Eindhoven University of Technology, Medical Electrical Engineering Group, The Netherlands.
Computer Methods and Programs in Biomedicine
|December 1, 1999
Summary
This study developed objective methods for detecting artefacts in intensive care unit (ICU) electroencephalogram (EEG) recordings. Combined autoregressive (AR) and Slope detection accurately identified about 90% of EEG artefacts.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Reliable artefact detection in electroencephalogram (EEG) is crucial for accurate analysis.
- Existing EEG analysis systems lack broadly applicable artefact recognition techniques.
- Intensive care unit (ICU) EEG recordings present unique challenges and opportunities for artefact detection.
Purpose of the Study:
- To investigate objective artefact detection methods for 24-hour ICU EEG recordings.
- To evaluate the performance of time-varying autoregressive (AR) modelling and Slope detection.
- To determine optimal settings for context incorporation in artefact detection algorithms.
Main Methods:
- Utilized seven 24-hour EEG recordings from an explorative ICU study.
- Applied statistical differences between signal parameters using time-varying AR modelling and Slope detection.
- Tested algorithms with varying time windows and epoch lengths, comparing against human observers.
Main Results:
- Objective methods based on statistical differences showed promise for EEG artefact detection.
- A relatively short context period (20-40 seconds) was sufficient for effective artefact detection.
- Combined AR and Slope detection parameters achieved approximately 90% artefact detection accuracy, matching human consensus.
Conclusions:
- Objective artefact detection methods, specifically combined AR and Slope detection, are effective for ICU EEG data.
- The findings support the use of these methods for improving the reliability of EEG analysis in critical care settings.
- Optimized algorithms with short context windows offer a practical solution for automated EEG artefact identification.