Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 9, 454

Automated Long-Term EEG Review: Fast and Precise Analysis in Critical Care Patients


Automated Long-Term EEG Review: Fast and Precise Analysis in Critical Care Patients

Johannes P Koren et al. Front Neurol.


Background: Ongoing or recurrent seizure activity without prominent motor features is a common burden in neurological critical care patients and people with epilepsy during ICU stays. Continuous EEG (CEEG) is the gold standard for detecting ongoing ictal EEG patterns and monitoring functional brain activity. However CEEG review is very demanding and time consuming. The purpose of the present multirater, EEG expert reviewer study, is to test and assess the clinical feasibility of an automatic EEG pattern detection method (Neurotrend). Methods: Four board certified EEG reviewers used Neurotrend to annotate 76 CEEG datasets à 6 h (in total 456 h of EEG) for rhythmic and periodic EEG patterns (RPP), unequivocal ictal EEG patterns and burst suppression. All reviewers had a predefined time limit of 5 min (± 2 min) per CEEG dataset and were compared to a predefined gold standard (conventional EEG review with unlimited time). Subanalysis of specific features of RPP was conducted as well. We used Gwet's AC1 and AC2 coefficients to calculate interrater agreement (IRA) and multirater agreement (MRA). Also, we determined individual performance measures for unequivocal ictal EEG patterns and burst suppression. Bonferroni-Holmes correction for multiple testing was applied to all statistical tests. Results: Mean review time was 3.3 min (± 1.9 min) per CEEG dataset. We found substantial IRA for unequivocal ictal EEG patterns (0.61-0.79; mean sensitivity 86.8%; mean specificity 82.2%, p < 0.001) and burst suppression (0.68-0.71; mean sensitivity 96.7%; mean specificity 76.9% p < 0.001). Two reviewers showed substantial IRA for RPP (0.68-0.72), whereas the other two showed moderate agreement (0.45-0.54), compared to the gold standard (p < 0.001). MRA showed almost perfect agreement for burst suppression (0.86) and moderate agreement for RPP (0.54) and unequivocal ictal EEG patterns (0.57). Conclusions: We demonstrated the clinical feasibility of an automatic critical care EEG pattern detection method on two levels: (1) reasonable high agreement compared to the gold standard, (2) reasonable short review times compared to previously reported EEG review times with conventional EEG analysis.

Keywords: continuous EEG; intensive care unit; neurotrend; non-convulsive seizures; standardized critical care EEG terminology; status epilepticus.


Figure 1
Figure 1
Overview of the NeuroTrend graphical user interface (GUI). (A) Automatic, color coded pattern detection (light blue: PD, periodic discharges; violet: RDA, rhythmic delta activity; pink: RDA+S, rhythmic delta activity plus superimposed spikes; orange: RTA, rhythmic theta activity; light green: RAA, rhythmic alpha activity); (B) Related frequencies of detected EEG patterns (the same color code as in A is used); (C) Amplitude integrated EEG for left and right hemisphere; (D) Frequency bands (beta-alpha-theta-delta) in a color coded (blue: beta; green: alpha; orange: theta; violet: delta), stacked proportion view (stronger colors signal higher amplitudes); (E) Burst suppression detection (continuous red markers signal presence of burst suppression); (F) Heart rate frequency plot. The black arrow highlights an EEG example of 1.5-2 c/s left hemispheric periodic discharges with superimposed rhythmic activity, which can be easily detected with the Neurotrend GUI.
Figure 2
Figure 2
Interpretation of NeuroTrend. (A) Recurrent seizures are detected as generalized rhythmic theta activity (RTA, orange plots) between 22:30 and 00:00. Then ongoing seizure activity is displayed by ongoing detection of RTA until 01:30. Around 01:00 detection of generalized rhythmic delta activity (RDA, pink and violet plots) overlap with RTA and further increases until 03:00. (B) Related pattern frequency detection reveal clear cut seizures above 3 c/s between 22:30 and 01:30 (black arrow). Overlapping RDA show a steady decrease from 3.5 to 2 c/s (red arrow). (C) Amplitude integrated EEG shows increment and decrement over both hemispheres at the beginning of each seizure from 22:30 to 23:30. Then a steady increase over both hemispheres can be seen during ongoing seizure activity from 00:00 until 01:00. (D) Frequency bands show a dominance of theta activity during seizure activity and the overlap of theta and delta activity around 01:30. (E) No burst suppression was detected. (F) Heart rate does not really show a concordance to seizure activity. In synopsis, this example represent typical spatiotemporal evolution of electrographic seizure activity, which can be easily detected with the graphical user interface of Neurotrend.
Figure 3
Figure 3
NeuroTrend example of a 49-year-old man with left temporal gliosis and sepsis. Six hours of continuous EEG (CEEG) are depicted with the Neurotrend GUI. Suppressed EEG due to sedoanalgesia can be clearly identified (black arrow). (A) No rhythmic or periodic EEG pattern was detected. (B) No pattern frequencies are displayed. (C) Amplitude integrated EEG shows a stable amplitude of 5–10 μV over both hemispheres. (D) Frequency bands show a low amplitude beta activity with underlying, low amplitude delta activity. (E) Burst suppression detection shows several periods with burst suppression. GUI, graphical user interface.
Figure 4
Figure 4
NeuroTrend examples of a 41-year-old woman with morphine abuse and sepsis. Six hours of continuous EEG (CEEG) are depicted with a compressed Neurotrend GUI in the top section (Amplitude integrated EEG, frequency bands, burst suppression detection, and heart rate frequency plot are hidden in this example). (A,B) display a stable detection of 1.5 c/s generalized rhythmic delta activity (GRDA, black arrow). The following 6 h of CEEG in the section below, show an overlap with a more periodic EEG pattern around 1 c/s after 3 h of recording (C,D, red arrow). GUI, graphical user interface.

Comment in

Similar articles

See all similar articles

Cited by 2 articles


    1. Young GB, Jordan KG, Doig GS. An assessment of nonconvulsive seizures in the intensive care unit using continuous EEG monitoring: an investigation of variables associated with mortality. Neurology (1996) 47:83–9. - PubMed
    1. Kaplan PW. Assessing the outcomes in patients with nonconvulsive status epilepticus: nonconvulsive status epilepticus is underdiagnosed, potentially overtreated, and confounded by comorbidity. J Clin Neurophysiol. (1999) 16:341–52. - PubMed
    1. Sutter R, Semmlack S, Kaplan PW. Nonconvulsive status epilepticus in adults - insights into the invisible. Nat Rev Neurol. (2016) 12:281–93. 10.1038/nrneurol.2016.45 - DOI - PubMed
    1. Sutter R, Kaplan PW. Electroencephalographic criteria for nonconvulsive status epilepticus: synopsis and comprehensive survey. Epilepsia (2012) 53(Suppl. 3):1–51. 10.1111/j.1528-1167.2012.03593.x - DOI - PubMed
    1. Claassen J, Perotte A, Albers D, Kleinberg S, Schmidt JM, Tu B, et al. . Nonconvulsive seizures after subarachnoid hemorrhage: multimodal detection and outcomes. Ann Neurol. (2013) 74:53–64. 10.1002/ana.23859 - DOI - PMC - PubMed

LinkOut - more resources