Consistency of quantitative electroencephalography features in a large clinical data set
David O Nahmias1,2, Kimberly L Kontson1, David A Soltysik1
1Division of Biomedical Physics, Office of Science and Engineering Laboratories, Center for Devices and Radiological Health, U.S. Food and Drug Administration, Silver Spring, Maryland, United States of America.
Journal of Neural Engineering
|October 5, 2019
Summary
This study introduces a new method to assess electroencephalography (EEG) feature consistency. Certain EEG metrics are reliable for baseline measurements, while others better distinguish individuals.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Quantitative electroencephalography (EEG) features are increasingly used in research and clinical settings.
- Understanding the consistency and variability of baseline EEG measurements in healthy populations is crucial but limited.
- This variability impacts the reliability and interpretation of EEG data across different individuals and studies.
Purpose of the Study:
- To investigate the population consistency and variability of quantitative features derived from baseline electroencephalography (EEG) measurements.
- To establish data-driven methods for identifying suitable EEG features for various applications.
- To determine the consistency of EEG features for future research using existing datasets.
Main Methods:
- A non-parametric method was developed to evaluate the consistency of commonly used EEG features.
- The method utilized counts of non-significant statistical tests on a large dataset.
- Evaluated feature stationarity, intra-subject consistency, inter-subject consistency, and intra- versus inter-subject consistency for 30 features across different epoch lengths.
Main Results:
- Features incorporating normalizing constants generally exhibited greater stationarity.
- Entropy, median, skew, and kurtosis of EEG were identified as reliable baseline EEG metrics.
- Other spectral and signal shape features demonstrated stronger intra-subject consistency, making them suitable for individual identification.
Conclusions:
- The study provides a robust, data-driven framework for assessing EEG feature consistency.
- The findings guide the selection of appropriate EEG features based on their statistical properties and intended application.
- This research contributes to a better understanding of EEG signal variability and its implications for clinical and research use.
