Abstracts

Rationale: Ripples, or high frequency oscillations in the ripple band (80-250 Hz) have been postulated as biomarkers of the seizure onset zone (SOZ). However, the presence of physiologic ripples in non-epileptic brain regions compromises their specificity. To characterize these physiologic events we investigated the rate, amplitude, and spatial distribution of ripples in non-epileptic brain regions. Methods: Channels in the SOZ and healthy channels were selected in 12 patients with intracerebral electrodes. Healthy channels were defined as channels showing no interictal epileptic activity, not in the SOZ, and outside any lesion visible in MRI. A maximum of two channels per electrode (most mesial and most lateral bipolar contacts) were selected to minimize the correlation between channels. An automatic detector was used since the expected low ripple rates in healthy channels require long intervals for their accurate estimation. The complete NREM (N2 and N3 stages) period of the first sleep cycle was investigated. To minimize false positives, we excluded channels with high amplitude background above 200 Hz and channels with a large variability in the background level. We also studied the spatial distribution of ripples in non-epileptic brain regions. A larger number of patients was investigated to obtain good coverage of many brain areas. The procedure described above was followed to estimate the ripple rate in healthy channels of 34 patients. Results: The mean duration of the N2/N3 stages of the first sleep cycle was 63 min. The median ripple rate in 84 healthy channels was 1.3 per min, and in 123 channels in SOZ, it was 4.7 per min. Using the ripple rate to discriminate between healthy channels and SOZ, the maximum accuracy (75%) was obtained for a threshold of 1.8 ripples per min, with 85% sensitivity and 60% specificity. The median amplitude of ripples in healthy channels (9.7 µV) was lower than in SOZ channels (12.6 µV). Figure 1 shows the obtained rates and amplitudes. Regarding the spatial distribution of ripples in non-epileptic brain regions, 240 channels from 34 patients were analyzed. Higher rates were observed in mesio-temporal structures, the precentral and central gyrus, and parieto-occipital regions (superior parietal lobule, posterior cingulate gyrus, precuneus, cuneus). Lower rates were observed for the remaining frontal lobe, insula, the inferior parietal lobule, and temporal neocortex. Figure 2 shows the rates and number of channels in each investigated region. Conclusions: The ripple rates in non-epileptic brain regions, presumably physiologic ripples, were as expected lower than in the SOZ. A reasonable separation between SOZ and non-epileptic regions could be obtained based on the ripple rates. However, there are considerable differences in ripple rates across the non-epileptic brain regions which could be used to improve the SOZ localization using region-specific ripple rates.