Abstracts

Rationale: We investigated the relationship between locations of broadband interictal electrocorticographic activities determined by the semi-automatic maximum wavecluster detection method and seizure outcomes in medically intractable epilepsy patients.Methods: Included were 22 patients with medically intractable epilepsy who underwent chronic electrocorticography. Seizure outcomes were followed more than 1 year. Investigated interictal periods had 20s durations and spikes with or without slowings. Prominent electrocorticographic activities were extracted in the wavelet transformed domain in 28 scales from 600Hz to 0.07Hz separately. Wavecluster, a continuous activity in wavelet transformed domain, was detected by thresholdings. Subsequently we selected one electrode having the maximum wavecluster amplitude in each seizure. Thereafter we compared the mean resection percentage of maximum electrodes between seizure outcome groups between the seizure-free and the not-seizure-free groups by Mann-Whitney U tests. Benjamini-Hochberg false discovery rate corrections were done for 28 scales. Significances were judged at the adjusted p-value of 0.05 level. Thresholds were automatically optimized with the genetic algorithm to find the greatest difference between seizure outcome groups.Results: Resection extents of maximum amplitude waveclusters were significantly different between seizure outcome groups in 6 scales including 2Hz, 1Hz, 0.7Hz, 0.5Hz, 0.13Hz and 0.09Hz scales (adjusted p values = 0.024, 0.031, 0.024, 0.024, 0.031 and 0.024). Fast ripples in 307 and 431 Hz scales were significantly different before multiple comparison corrections (unadjusted p = 0.031 and 0.022). However, after multiple comparison corrections, there were not significantly related with seizure outcomes.Conclusions: The significant relationship between resection extents of zones with slow oscillations and seizure outcomes was found in the interictal electrocorticography. The automatic analysis of short 20 duration of interictal period with abnormal activities were sufficient to find differences between differences between seizure outcome groups. Better delineation of resection targets may be possible in future by broadband electrocorticography.