Related Experiment Video
Updated: Aug 20, 2026
Application of an Amplitude-integrated EEG Monitor (Cerebral Function Monitor) to Neonates
Published on: September 6, 2017
Neonatal seizure monitoring using non-linear EEG analysis
L S Smit1, R J Vermeulen, W P F Fetter
1Department of Clinical Neurophysiology/MEG Centre, VU University Medical Centre, Amsterdam, The Netherlands.
Insights
Synchronization likelihood analysis effectively detects subclinical neonatal seizures from raw EEG data. This method shows promise for automated monitoring in neonatal intensive care units, improving infant seizure detection rates.
Area of Science:
- Neonatal neurology
- Computational neuroscience
- Medical device development
Background:
- Birth asphyxia is a significant cause of neonatal complications.
- Subclinical epileptic seizures in neonates can lead to neurodevelopmental deficits.
- Current monitoring methods lack the capacity to detect the majority of these subclinical seizures.
Purpose of the Study:
- To evaluate the efficacy of synchronization likelihood analysis in detecting neonatal epileptic seizures using complete, unfiltered EEG data.
- To assess the performance of synchronization likelihood in identifying seizure activity without prior artifact removal.
- To determine the impact of seizure duration on detection rates.
Main Methods:
- Analysis of 20 complete neonatal EEG recordings from 20 patients.
- Calculation of synchronization likelihood on raw, unfiltered EEG data.
- Correlation of synchronization likelihood results with expert visual scoring and assessment of seizure detection rates based on seizure length.
Main Results:
- Sensitivity of 65.9% and specificity of 89.8% for epoch-based seizure detection using raw EEG.
- 100% seizure detection rate for seizures lasting 100 seconds or longer.
- Synchronization likelihood demonstrates potential for distinguishing between seizure and non-seizure epochs.
Conclusions:
- Synchronization likelihood is a viable tool for automated monitoring of neonatal epileptic seizures.
- Further prospective studies are required to establish clinical intervention consequences.
- Development of an on-line version of the analysis will be pursued for real-time application in neonatal intensive care units.
Abstract:
Birth asphyxia is a major concern in neonatal care. Epileptic seizures are associated with subsequent neurodevelopmental deficits. Eighty-five percent of these seizures remain subclinical and therefore an on-line monitoring device is needed. In an earlier study we showed that the synchronization likelihood was able to distinguish between neonatal EEG epochs with and without epileptic seizures. In this study we investigated whether the synchronization likelihood can be used in complete EEGs, without artifact removal. Twenty complete EEGs from 20 neonatal patients were studied. The synchronization likelihood was calculated and correlated with the visual scoring done by 3 experts. In addition, we determined the influence of seizure length on the likelihood of detection. Using the raw unfiltered EEG data we found a sensitivity of 65.9 % and a specificity of 89.8 % for the detection of seizure activity in each epoch. In addition, the seizure detection rate was 100 % when the seizures lasted for 100 seconds or more. The synchronization likelihood seems to be a useful tool in the automatic monitoring of epileptic seizures in infants on the neonatal ward. Due to the retrospective nature of our study, the consequences for clinical intervention cannot yet be determined and prospective studies are needed. Therefore, we will conduct a prospective study on the neonatal intensive care unit with a recently developed on-line version of the synchronization likelihood analysis.
