Abstracts

Rationale: EEG digital trend analysis (DTA) presents EEG waveforms in a graphic format that compress hours of EEG data into a single display. Spectral analysis (SA) displays power versus frequency versus time in a three dimensional view whereas envelop trend (ET) is a time averaging technique, developed especially to detect seizures. We compared these two techniques to determine which is a better compliment to raw EEG data for accurate seizure detection in children. Methods: The EEG-video recordings of 11 children (ages 3 days to 14 years) urgently admitted for seizures were examined. 151 seizures were identified on raw EEG; clinical and EEG features were examined, including the amplitude of EEG background and seizures; SA and ET were applied to each seizure, both independently and then together. Only CIA reviewed the raw EEG and DTA (SA and ET) data. Two DTA size displays were prepared. DTA data alone (without raw EEG) were presented to both an experienced (JR) and an inexperienced user (VM), first individually and then together. Based on these results, the sensitivity for each DTA technique was assessed. Results: 114/151 (76%) seizures were non-convulsive seizures (NCSz). The duration of seizures varied between 20-360 seconds. Except for four seizures, all seizures were focal, 64 arising form left and 83 from the right hemisphere. Seven seizures occurring with even intervals every 2 minutes were classified as cyclic seizures. For all 151 seizures, 26 tracings were produced using the 2 display sizes. In one patient, ET failed to identify seizures whereas SA clearly demonstrated the seizures; this was related to EEG background amplitude. In three patients, ET alone demonstrated seizures, which were not visible with SA. With the initial display size using SA, seizure detection sensitivity was 53% by both an experienced and inexperienced user. Doubling the display size increased the SA sensitivity to 80% with the experienced and 69% with the inexperienced user. When ET technique was used alone for interpretation, sensitivity for seizure detection was 84% with the experienced user, 57% with the inexperienced user. With both techniques combined, the experienced user missed no seizure; the inexperienced user missed only 5 seizures. Only the experienced user identified cyclic seizures in 7 tracings by using SA alone compared to only 3 tracings with ET alone. Conclusions: Multimodality DTA combining SA and ET increases the accuracy of seizure detection, particularly for cyclic seizures in children. DTA also increases the detection of NCSz. EEG background amplitude and display size are important factors affecting the sensitivity of seizure detection.