Abstracts

Rationale: Scalp EEG-video monitoring is a standard procedure in the pre-surgical evaluation of patients suffering from medically intractable epileptic seizures. The efficiency of this time and labor intensive task depends on the timeliness and accuracy with which recorded seizures can be identified in multi-day recordings, especially those lacking clear behavior alterations. With advances in signal processing techniques and computing power, several computer software systems offering automated seizure detection (ASD) have been utilized in epilepsy monitoring units (EMUs). Patients in many EMUs include both children and adults. Since their EEG characteristics (of both ictal discharges and background activities) are well documented to differ in some degree, the performance of ASD algorithms for EMUs should be evaluated for not only adult patients, but pediatric patients as well. We have previously reported the performance of our EMU seizure detection software (IdentEvent ) in adults (age 18 and above). The objective of this study is to evaluate the performance of the IdentEvent detection algorithm specifically in pediatric (age 3 17) EMU patients.Methods: Twenty-two prolonged cEEG recordings from 20 pediatric EMU patients were studied (total duration 1,772 hours with a total of 98 seizures). Sixteen recordings contained at least one complex partial (with or without generalization), myoclonic, or primary generalized seizure. The study excluded recordings with frequent brief primary generalized absence seizures. For each subject, the entire EEG recording (no clipping) was included in the analysis to create the most representative clinical data set possible by including patients normal physiological conditions and recording artifacts. Using quantitative EEG (qEEG) measurements of signal regularity, power, and frequency (calculated for each 5s epoch), the IdentEvent ASD algorithm detects significant changes of qEEGs from the preceding baseline means as well as the asymmetry between hemispheres. The algorithm also utilized qEEGs to reject artifacts such as recording noise, movement/muscle, and electrode failure. We evaluated the performance of the algorithm by estimating the detection sensitivity and the false detection rate (per 24hrs).Results: Overall, the IdentEvent algorithm accurately detected 86 seizures (87.8% sensitivity) with a mean false detection rate of 0.17/hr (or 4.2/day). Compared to its performance in adult EMU patients (~90% with 3.0 false detections/day), the sensitivities are similar, but it generates more false detections in pediatric patients, especially for those with more abnormal background EEGs.Conclusions: The results from this study suggest that the IdentEvent ASD algorithm performs well on pediatric EMU patients and therefore can be useful clinically in enhancing the efficiency of scalp video-EEG monitoring for all age ( 3) patient groups in the EMUs. User-adjustable detection settings for patients with particularly abnormal background EEGs, which may be observable shortly after the recording starts, could further enhance the usability of the ASD software.