Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2010 Jul;31(7):1047-64.
doi: 10.1088/0967-3334/31/7/013. Epub 2010 Jun 28.

Gaussian mixture models for classification of neonatal seizures using EEG

E M Thomas  1 , A Temko, G Lightbody, W P Marnane, G B Boylan
Affiliations

Gaussian mixture models for classification of neonatal seizures using EEG

E M Thomas et al. Physiol Meas. 2010 Jul.

Abstract

A real-time neonatal seizure detection system is proposed based on a Gaussian mixture model classifier. The system includes feature transformation techniques and classifier output postprocessing. The detector was evaluated on a database of 20 patients with 330 h of recordings. A detailed analysis of the choice of parameters for the detector is provided. A mean good detection rate of 79% was obtained with only 0.5 false detections per hour. A thorough review of all misclassified events was performed, from which a number of patterns causing false detections were identified.

PubMed Disclaimer

Figures

Figure 1
Figure 1
Example of two seizure patterns for a single patient. Here, the high amplitude seizure is generalised and manifests with a lower fundamental frequency. In contrast, the low amplitude seizure is localised to channels connected to C4 and has a higher fundamental frequency than the high amplitude seizure.
Figure 2
Figure 2
Block diagram of the neonatal seizure detector.
Figure 3
Figure 3
Diagram of the GMM based classifier.
Figure 4
Figure 4
Example of the effects of postprocessing. The ground truth is shown as dashed lines, with 1 indicating a seizure. Plot a) shows the probability of seizure prior to postprocessing. Plot b) shows the resulting binary decisions when postprocessing is not used. It should be noted that short transients cause a large number of false detections. In plot c), the output of the 15 point moving average filter is shown. Plot d) shows the binary decisions resulting from the filtered probability of seizure, note that false detections have been reduced to 1 false detection seen at 10 minutes. Plot e) shows the final binary decisions after the collar operation, which increases the duration of all positive decisions. This results in higher sensitivity, but also increases the duration of the false detection.
Figure 5
Figure 5
The confusion matrix used to compute the epoch based metrics and the difference between epoch and event based metric calculations.
Figure 6
Figure 6
ROC area as a function of moving average and collar widths.
Figure 7
Figure 7
(a) Mean ROC curve and (b) mean GDR as a function of FD/h. These curves are obtained over the LOO set and the with 95% confidence interval is shown over all patients in the set. Also included are markers showing the reported performance from recent algorithms.
Figure 8
Figure 8
Examples of false detections occurring from artefact-free EEG.
Figure 9
Figure 9
Examples of the most common artefacts leading to false detections.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Aarabi A, Grebe R, Wallois F. A multistage knowledge-based system for EEG seizure detection in newborn infants. Clin. Neurophysiol. 2007;118(no. 12):2781–97. - PubMed
    1. Aarabi A, Wallois F, Grebe R. Automated neonatal seizure detection, a multistage classification system through feature selection based on relevance and redundancy analysis. Clin. Neurophysiol. 2006;117(no. 2):328–40. - PubMed
    1. Bye AME, Flanagan D. Spatial and temporal characteristics of neonatal seizures. Epilepsia. 1995;36(no. 10):1009–16. - PubMed
    1. Celka P, Colditz P. A computer-aided detection of EEG seizures in infants, a singular-spectrum approach and performance comparison. IEEE Trans. Biomed. Eng. 2002;49(no. 5):455–462. - PubMed
    1. Clancy R. Summary proceedings from the neurology group on neonatal seizures. Pediatrics. 2006;117:23–27. - PubMed

Publication types

  NODES
Note 2
twitter 2