Abstracts

Rationale: EEG interpretation can be impaired by artifact. A new type of tripolar concentric ring electrode (TCRE) comprises 3 circular recording zones with a differential (subtractive) output that cancels artifact common to all 3 zones. This study compares EEG recorded with tripolar versus regular electrodes. Methods: EEG segments clipped to 30 seconds durations, containing seizure onsets, spikes, focal slowing and normal waking or sleeping waveforms. EEG was recorded simultaneously by TCRE and adjacent regular (REG) electrodes. Segments of EEG were chosen non-randomly to include EEG potentials of interest. REG EEG was filtered with high cutoff at 35 Hz (REG35), at 70 Hz (REG70) and TCRE at 70 Hz. Clip order was randomized, and presented via Survey Monkey, which did not allow return to prior screens or answers. EEG readers (n=9) blindly selected from multiple choices whether various EEG waveforms, seizures and artifacts were present, the overall quality of the recording and confidence of interpretation (1-5, with 1 best). Results: Averaged responses from raters were included in the analysis. Cronbach’s alpha showed medium-high reliability on most measures. A One-Way ANOVA suggested a significant difference between our three groups, F >48 (p < .001). Bonferroni post-hocs revealed that TCRE electrodes were rated to have less EMG (M = 1.06 ± 0.88), compared to REG35 (1.31±1.06, t >-4.57, p < .001) and REG70 (1.59 ± 1.59, t >-9.47 p < .001). TCRE was rated worse for sweat and POP artifact, and equivalent for movement and amplifier blocking, and in confidence of interpretation. Overall quality rankings were similar for all 3 electrodes. Among 85 rated individual clips containing seizures, at least one of the electrodes allowed detection of a seizure in 82 (96.5%) of seizures. Seizures were detected with 64 (75.3%) TCRE, 75 (88.2%) REG70 and 69 (81.2%) REG35 electrodes. Concordance for all 3 electrodes occurred with 55 (64.7%) seizures. TCRE detected 9 (10.6%) seizures not detected by one of the other 2 methods and 6 (7.1%) not detected by either. Figure 1 shows a seizure detected only by TCRE. In contrast, 14 (16.5%) seizures were not detected by TCRE, but were by REG35 electrodes. TCRE occasionally detected interictal spikes when the other electrodes did not and vice versa. Conclusions: TCRE produced higher blinded ratings for overall quality versus regular electrodes filtered at 35 Hz and consistently demonstrated less muscle artifact. Other artifacts were not reduced. TCRE recordings detected seizures in 7% of instances where regular electrodes did not. However, the reverse was true for 16.5% of seizures detected only by regular electrodes. TCRE recording is not intrinsically superior to regular filtered recordings, but the combination of the two (TCRE electrodes also can output an equivalent of regular electrode signals) increases detection of epileptiform events compared to either alone. Funding: RSF is supported by the Maslah Saul Chair, the Anderson, Chen, Horngren funds for epilepsy research and NSF grant 1430833. Conflicts: RSF has no conflicts relevant to this abstract. Walter Besio is an employee of CREmedical Corporation, manufacturer of the TCRE.