• Ajay Goenka, Alexis Boro, and Elissa Yozawitz.
• Saul Korey Department of Neurology, Montefiore Medical Center and Albert Einstein College of Medicine, Bronx, NY 10467, United States. Electronic address: goenkaa@childrensdayton.org.
• Seizure. 2018 Feb 1; 55: 70-75.

PurposeTo assess the sensitivity of Persyst version 12 QEEG spectrograms to detect focal, focal with secondarily generalized, and generalized onset seizures.MethodsA cohort of 562 seizures from 58 patients was analyzed. Successive recordings with 2 or more seizures during continuous EEG monitoring for clinical indications in the ICU or EMU between July 2016 and January 2017 were included. Patient ages ranged from 5 to 64 years (mean = 36 years). There were 125 focal seizures, 187 secondarily generalized and 250 generalized seizures from 58 patients analyzed. Seizures were identified and classified independently by two epileptologists. A correlate to the seizure pattern in the raw EEG was sought in the QEEG spectrograms in 4-6 h EEG epochs surrounding the identified seizures. A given spectrogram was interpreted as indicating a seizure, if at the time of a seizure it showed a visually significant departure from the pre-event baseline. Sensitivities for seizure detection using each spectrogram were determined for each seizure subtype.ResultsOverall sensitivities of the QEEG spectrograms for detecting seizures ranged from 43% to 72%, with highest sensitivity (402/562,72%) by the seizure detection trend. The asymmetry spectrogram had the highest sensitivity for detecting focal seizures (117/125,94%). The FFT spectrogram was most sensitive for detecting secondarily generalized seizures (158/187, 84%). The seizure detection trend was the most sensitive for generalized onset seizures (197/250,79%).ConclusionsOur study suggests that different seizure types have specific patterns in the Persyst QEEG spectrograms. Identifying these patterns in the EEG can significantly increase the sensitivity for seizure identification.Copyright © 2018 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.

23 keywords...

hide…