Abstracts

Rationale: High frequency oscillations (HFOs) between 80-500Hz can identify the seizure onset zone (SOZ) with high specificity. Spikes with distinct HFOs are more specific for the SOZ than spikes without HFOs. Visual or semi-automatic detection of HFOs however can be time consuming and difficult. Another method uses statistical analysis of averaged power spectra of interictal spikes to identify those spikes with large high frequency (HF) components. This study compares the value of distinct HFOs and HF components of spikes for identifying the SOZ. Methods: 15 patients with neocortical refractory epilepsy from the Montreal Neurological Institute were included (low pass: 500Hz, sampling rate: 2000Hz). HFOs inside and separate of spikes were marked visually and rates of HFOs (ripple: 80-250Hz, fast ripple: 250-500Hz) were calculated. For the analysis of HF components, at least 50 spikes with similar morphologies were marked in all spiking channels. Time-frequency spectra of spikes were averaged and FDR corrected statistics was used to identify significant increases and decreases of frequencies between 80-250Hz as well as 250-500Hz (Figure 1). Specificity and sensitivity for identifying the SOZ were calculated on a single patient and group level and compared for both methods. Results: In 71 channels (16 in the SOZ), polyspikes were observed which did not allow the analysis of HF power within these spikes. A significantly higher rate of HFOs and spikes carrying HFOs was found in the SOZ compared to outside. At a level of 95% specificity the rate of spikes carrying ripples was the most sensitive (30%) to identify the SOZ. When applying the rate threshold at the 95% specificity level to single patients, either spikes carrying ripple or spikes carrying fast ripple showed the highest sensitivity in 11 and the highest specificity in 14 of 15 patients. For the increase of HF power within spikes no difference was found inside or outside the SOZ whereas there was a significantly larger postspike decrease for frequencies between 80-250Hz and 250-500Hz inside the SOZ (p=0.01). Thus, HF power increases in the spikes could not be used for identification of the SOZ. Conclusions: In patients with neocortical epilepsy, spikes which were either carrying ripple or fast ripples were best for identifying the SOZ. Thus, it could be expected that HF power in spikes, if similar to distinct HFOs, would show the same discriminatory features. In contrast, only the postspike decrease in HF power was different for the SOZ. Thus, strong HF power changes in spikes and spikes carrying HFOs seem to represent different entities and may occur over different regions. This may be additionally influenced by methodological problems occurring in channels with very frequent spiking, which prevent HF power analysis in spikes. The observed prominent postspike decrease in HF power in SOZ channels may represent increase inhibition in these channels and should be investigated further.