Related Experiment Video
Updated: Jul 16, 2025
05:58
Application of an Amplitude-integrated EEG Monitor Cerebral Function Monitor to Neonates
Published on: September 6, 2017
38.9K
Quantitative analysis of high-frequency activity in neonatal EEG
Christopher Lundy1, Geraldine B Boylan1, Sean Mathieson1
1INFANT Research Centre, University College Cork, Cork, Ireland; Department of Paediatrics and Child Health, University College Cork, Cork, Ireland.
Computers in Biology and Medicine
|September 18, 2023
Summary
High-frequency electroencephalogram (EEG) activity in newborns, particularly preterms, shows independent utility for analysis. Expanding EEG bandwidths improves quantitative and machine learning approaches for neonatal brain activity assessment.
Area of Science:
- Neonatal neurology
- Neurophysiology
- Biomedical signal processing
Background:
- Electroencephalogram (EEG) is crucial for assessing neonatal brain function.
- Traditional EEG analysis often focuses on standard frequency bands, potentially overlooking valuable information in higher frequencies.
- Understanding high-frequency EEG activity may enhance diagnostic capabilities in vulnerable infant populations.
Purpose of the Study:
- To investigate the presence and diagnostic utility of independent high-frequency activity in newborn EEG.
- To compare high-frequency EEG features with standard frequency bands in different newborn groups.
- To evaluate the performance of machine learning models utilizing high-frequency EEG data.
Main Methods:
- Retrospective analysis of 256 Hz EEG data from four newborn groups: preterm (<32 weeks GA), late preterm (32-37 weeks GA), healthy term, and term with hypoxic-ischemic encephalopathy (HIE).
- Comparison of power spectral densities (PSDs) and quantitative features across standard (delta, theta, alpha, beta) and high-frequency bands (gamma1, gamma2, gamma3).
- Application and evaluation of machine learning models, including feature selection, on EEG data.
Main Results:
- Significant differences in PSDs and quantitative analysis were observed in high-frequency bands (P < 0.01).
- Machine learning models using solely high-frequency features demonstrated strong performance in preterm groups (MCC 0.71 and 0.66).
- High-frequency features were largely independent of standard-bandwidth features, contributing unique information.
Conclusions:
- This study is the first to identify independent high-frequency activity in newborn EEG through quantitative analysis.
- Expanding EEG analysis bandwidths can significantly improve quantitative and machine learning assessments, especially for preterm infants.
- High-frequency EEG analysis holds promise for enhanced neurophysiological evaluation in newborns.
