Assessing quantitative EEG spectrograms to identify non-epileptic events

Research output: Contribution to journalArticle

Abstract

Aims. To evaluate the sensitivity and specificity of quantitative EEG (QEEG) spectrograms in order to distinguish epileptic from non-epileptic events. Methods. Seventeen patients with paroxysmal non-epileptic events, captured during EEG monitoring, were retrospectively assessed using QEEG spectrograms. These patients were compared to a control group of 13 consecutive patients (ages 25-60 years) with epileptic seizures of similar semiology. Assessment of raw EEG was employed as the gold standard against which epileptic and non-epileptic events were validated. QEEG spectrograms, available using Persyst 12 EEG system integration software, were each assessed with respect to their usefulness to distinguish epileptic from non-epileptic seizures. The given spectrogram was interpreted as indicating a seizure if, at the time of the clinically identified event, it showed a visually significant change from baseline. Results. Eighty-two clinically identified paroxysmal events were analysed (46 non-epileptic and 36 epileptic). The “seizure detector trend analysis” spectrogram correctly classified 33/46 (71%) non-epileptic events (no seizure indicated during a clinically identified event) vs. 29/36 (81%) epileptic seizures (seizure indicated during a clinically identified event) (p=0.013). Similarly, “rhythmicity spectrogram”, FFT spectrogram, “asymmetry relative spectrogram”, and integrated-amplitude EEG spectrogram detected 28/46 (61%), 30/46 (65%), 22/46 (48%) and 27/46 (59%) non-epileptic events vs. 27/36 (75%), 25/36 (69%), 25/36 (69%) and 27/36 (75%) epileptic events, respectively. Conclusions. High sensitivities and specificities for QEEG seizure detection analyses suggest that QEEG may have a role at the bedside to facilitate early differentiation between epileptic seizures and non-epileptic events in order to avoid unnecessary administration of antiepileptic drugs and possible iatrogenic consequences.

Original languageEnglish (US)
Pages (from-to)299-306
Number of pages8
JournalEpileptic Disorders
Volume19
Issue number3
DOIs
StatePublished - Sep 1 2017

Fingerprint

Electroencephalography
Seizures
Epilepsy
Systems Integration
Sensitivity and Specificity
Periodicity
Anticonvulsants
Software
Control Groups

Keywords

  • jerking
  • PNES
  • psychogenic non-epileptic seizures
  • quantitative EEG
  • seizure detection trend
  • shaking

ASJC Scopus subject areas

  • Neurology
  • Clinical Neurology

Cite this

Assessing quantitative EEG spectrograms to identify non-epileptic events. / Goenka, Ajay; Boro, Alexis D.; Yozawitz, Elissa G.

In: Epileptic Disorders, Vol. 19, No. 3, 01.09.2017, p. 299-306.

Research output: Contribution to journalArticle

@article{5fef82eadfb84bc7aa5b05f37d65847d,
title = "Assessing quantitative EEG spectrograms to identify non-epileptic events",
abstract = "Aims. To evaluate the sensitivity and specificity of quantitative EEG (QEEG) spectrograms in order to distinguish epileptic from non-epileptic events. Methods. Seventeen patients with paroxysmal non-epileptic events, captured during EEG monitoring, were retrospectively assessed using QEEG spectrograms. These patients were compared to a control group of 13 consecutive patients (ages 25-60 years) with epileptic seizures of similar semiology. Assessment of raw EEG was employed as the gold standard against which epileptic and non-epileptic events were validated. QEEG spectrograms, available using Persyst 12 EEG system integration software, were each assessed with respect to their usefulness to distinguish epileptic from non-epileptic seizures. The given spectrogram was interpreted as indicating a seizure if, at the time of the clinically identified event, it showed a visually significant change from baseline. Results. Eighty-two clinically identified paroxysmal events were analysed (46 non-epileptic and 36 epileptic). The “seizure detector trend analysis” spectrogram correctly classified 33/46 (71{\%}) non-epileptic events (no seizure indicated during a clinically identified event) vs. 29/36 (81{\%}) epileptic seizures (seizure indicated during a clinically identified event) (p=0.013). Similarly, “rhythmicity spectrogram”, FFT spectrogram, “asymmetry relative spectrogram”, and integrated-amplitude EEG spectrogram detected 28/46 (61{\%}), 30/46 (65{\%}), 22/46 (48{\%}) and 27/46 (59{\%}) non-epileptic events vs. 27/36 (75{\%}), 25/36 (69{\%}), 25/36 (69{\%}) and 27/36 (75{\%}) epileptic events, respectively. Conclusions. High sensitivities and specificities for QEEG seizure detection analyses suggest that QEEG may have a role at the bedside to facilitate early differentiation between epileptic seizures and non-epileptic events in order to avoid unnecessary administration of antiepileptic drugs and possible iatrogenic consequences.",
keywords = "jerking, PNES, psychogenic non-epileptic seizures, quantitative EEG, seizure detection trend, shaking",
author = "Ajay Goenka and Boro, {Alexis D.} and Yozawitz, {Elissa G.}",
year = "2017",
month = "9",
day = "1",
doi = "10.1684/epd.2017.0921",
language = "English (US)",
volume = "19",
pages = "299--306",
journal = "Epileptic Disorders",
issn = "1294-9361",
publisher = "Springer Paris",
number = "3",

}

TY - JOUR

T1 - Assessing quantitative EEG spectrograms to identify non-epileptic events

AU - Goenka, Ajay

AU - Boro, Alexis D.

AU - Yozawitz, Elissa G.

PY - 2017/9/1

Y1 - 2017/9/1

N2 - Aims. To evaluate the sensitivity and specificity of quantitative EEG (QEEG) spectrograms in order to distinguish epileptic from non-epileptic events. Methods. Seventeen patients with paroxysmal non-epileptic events, captured during EEG monitoring, were retrospectively assessed using QEEG spectrograms. These patients were compared to a control group of 13 consecutive patients (ages 25-60 years) with epileptic seizures of similar semiology. Assessment of raw EEG was employed as the gold standard against which epileptic and non-epileptic events were validated. QEEG spectrograms, available using Persyst 12 EEG system integration software, were each assessed with respect to their usefulness to distinguish epileptic from non-epileptic seizures. The given spectrogram was interpreted as indicating a seizure if, at the time of the clinically identified event, it showed a visually significant change from baseline. Results. Eighty-two clinically identified paroxysmal events were analysed (46 non-epileptic and 36 epileptic). The “seizure detector trend analysis” spectrogram correctly classified 33/46 (71%) non-epileptic events (no seizure indicated during a clinically identified event) vs. 29/36 (81%) epileptic seizures (seizure indicated during a clinically identified event) (p=0.013). Similarly, “rhythmicity spectrogram”, FFT spectrogram, “asymmetry relative spectrogram”, and integrated-amplitude EEG spectrogram detected 28/46 (61%), 30/46 (65%), 22/46 (48%) and 27/46 (59%) non-epileptic events vs. 27/36 (75%), 25/36 (69%), 25/36 (69%) and 27/36 (75%) epileptic events, respectively. Conclusions. High sensitivities and specificities for QEEG seizure detection analyses suggest that QEEG may have a role at the bedside to facilitate early differentiation between epileptic seizures and non-epileptic events in order to avoid unnecessary administration of antiepileptic drugs and possible iatrogenic consequences.

AB - Aims. To evaluate the sensitivity and specificity of quantitative EEG (QEEG) spectrograms in order to distinguish epileptic from non-epileptic events. Methods. Seventeen patients with paroxysmal non-epileptic events, captured during EEG monitoring, were retrospectively assessed using QEEG spectrograms. These patients were compared to a control group of 13 consecutive patients (ages 25-60 years) with epileptic seizures of similar semiology. Assessment of raw EEG was employed as the gold standard against which epileptic and non-epileptic events were validated. QEEG spectrograms, available using Persyst 12 EEG system integration software, were each assessed with respect to their usefulness to distinguish epileptic from non-epileptic seizures. The given spectrogram was interpreted as indicating a seizure if, at the time of the clinically identified event, it showed a visually significant change from baseline. Results. Eighty-two clinically identified paroxysmal events were analysed (46 non-epileptic and 36 epileptic). The “seizure detector trend analysis” spectrogram correctly classified 33/46 (71%) non-epileptic events (no seizure indicated during a clinically identified event) vs. 29/36 (81%) epileptic seizures (seizure indicated during a clinically identified event) (p=0.013). Similarly, “rhythmicity spectrogram”, FFT spectrogram, “asymmetry relative spectrogram”, and integrated-amplitude EEG spectrogram detected 28/46 (61%), 30/46 (65%), 22/46 (48%) and 27/46 (59%) non-epileptic events vs. 27/36 (75%), 25/36 (69%), 25/36 (69%) and 27/36 (75%) epileptic events, respectively. Conclusions. High sensitivities and specificities for QEEG seizure detection analyses suggest that QEEG may have a role at the bedside to facilitate early differentiation between epileptic seizures and non-epileptic events in order to avoid unnecessary administration of antiepileptic drugs and possible iatrogenic consequences.

KW - jerking

KW - PNES

KW - psychogenic non-epileptic seizures

KW - quantitative EEG

KW - seizure detection trend

KW - shaking

UR - http://www.scopus.com/inward/record.url?scp=85033212128&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85033212128&partnerID=8YFLogxK

U2 - 10.1684/epd.2017.0921

DO - 10.1684/epd.2017.0921

M3 - Article

VL - 19

SP - 299

EP - 306

JO - Epileptic Disorders

JF - Epileptic Disorders

SN - 1294-9361

IS - 3

ER -