Related Experiment Video
Updated: Mar 3, 2026
Preterm EEG: A Multimodal Neurophysiological Protocol
Published on: February 18, 2012
Multi-feature classifiers for burst detection in single EEG channels from preterm infants
X Navarro1, F Porée2,3, M Kuchenbuch2,3,4
1Sorbonne Universités, UPMC Univ Paris 06, INSERM UMRS-1158 Neurophysiologie Respiratoire Expérimentale et Clinique, Paris, France.
Insights
This study developed a logistic regression (LR) model for detecting electroencephalographic (EEG) bursts in preterm infants nearing term age. The LR model accurately assesses infant brain maturation using single-channel EEG data.
Area of Science:
- Neonatal neurology
- Computational neuroscience
- Medical signal processing
Background:
- Electroencephalographic (EEG) bursts are crucial for assessing preterm infant maturation and prognostication after perinatal asphyxia.
- Existing automated EEG burst detection algorithms are primarily designed for infants under 35 weeks postmenstrual age (PMA), potentially underperforming in older infants.
- Brain activity evolves rapidly, necessitating specialized algorithms for preterm infants approaching term-equivalent ages (PMA ≥ 36 weeks).
Purpose of the Study:
- To develop and evaluate machine learning-based algorithms for automatic EEG burst detection in preterm infants with PMA ≥ 36 weeks.
- To compare the performance of different classifiers, including logistic regression (LR), linear discriminant analysis (LDA), k-nearest neighbors (kNN), support vector machines (SVM), and thresholding (Th).
- To identify features indicative of brain maturation in preterm infants using single-channel EEG data.
Main Methods:
- Five distinct machine learning classifiers (LR, LDA, kNN, SVM, Th) were implemented for EEG burst detection.
- Classifiers were trained and validated using visually labeled, single-channel EEG recordings from 14 preterm infants (born >28 weeks gestation) aged 36-41 weeks PMA.
- Performance was evaluated using accuracy and Cohen's kappa for inter-rater agreement.
Main Results:
- The kNN, SVM, and LR classifiers achieved approximately 95% accuracy, significantly outperforming thresholding (84%).
- Logistic regression, using only three EEG features, demonstrated the highest agreement with human experts (Cohen's kappa = 0.71).
- Application of the LR classifier to a larger dataset (21 infants ≥ 36 weeks PMA) revealed that longer EEG bursts and shorter inter-burst intervals correlate with higher PMA and weight.
Conclusions:
- Logistic regression-based EEG burst detection is a viable and accurate method for assessing maturation in preterm infants at term-equivalent ages.
- The developed LR model, utilizing single-channel EEG, offers a promising tool for monitoring brain development in clinical or portable devices.
- This approach can aid in understanding neurodevelopmental trajectories in preterm infants nearing term.
Objective:
The study of electroencephalographic (EEG) bursts in preterm infants provides valuable information about maturation or prognostication after perinatal asphyxia. Over the last two decades, a number of works proposed algorithms to automatically detect EEG bursts in preterm infants, but they were designed for populations under 35 weeks of post menstrual age (PMA). However, as the brain activity evolves rapidly during postnatal life, these solutions might be under-performing with increasing PMA. In this work we focused on preterm infants reaching term ages (PMA ⩾36 weeks) using multi-feature classification on a single EEG channel.
Approach:
Five EEG burst detectors relying on different machine learning approaches were compared: logistic regression (LR), linear discriminant analysis (LDA), k-nearest neighbors (kNN), support vector machines (SVM) and thresholding (Th). Classifiers were trained by visually labeled EEG recordings from 14 very preterm infants (born after 28 weeks of gestation) with 36-41 weeks PMA.
Main Results:
The most performing classifiers reached about 95% accuracy (kNN, SVM and LR) whereas Th obtained 84%. Compared to human-automatic agreements, LR provided the highest scores (Cohen's kappa = 0.71) using only three EEG features. Applying this classifier in an unlabeled database of 21 infants ⩾36 weeks PMA, we found that long EEG bursts and short inter-burst periods are characteristic of infants with the highest PMA and weights.
Significance:
In view of these results, LR-based burst detection could be a suitable tool to study maturation in monitoring or portable devices using a single EEG channel.
