Abstracts

Rationale: Identifying the networks underlying epileptic seizures remains an important step toward understanding the mechanisms behind seizure propagation. Here, we use a two-pronged approach using inter-ictal intracranial EEG recordings from human participants. Specifically, we quantify overrepresented serial inter-ictal spike sequences and construct functional connectivity maps for each participant in order to look for evidence of the spike-spike relationships in the functional networks. Methods: We identify inter-ictal spikes by finding biphasic waveforms which exceed a voltage threshold, and imposing further constraints on the duration of these waveforms. Spike sequences were built by considering 50ms windows which contained spikes from multiple electrodes and ordering the sequence based on the time of the first peak (or through depending on the polarity of the spike) for each spike. In order to judge the statistical significance of specific inter-ictal spike sequences we use a conditional probability based approach (Sastry and Unnikrishnan, Neural Computation 2010) which allows us to rank-order the individual spike sequences. The method by which we construct functional networks is based on time-lagged mutual information, and has been described in a previous study (Chapeton et al. Brain, 2017). Results: In most participants, we find statistically significant spike sequences, and the channels that participate in these sequences are in good agreement with the channels which were marked as inter-ictal by the clinical team. However, we also find several channels which have inter-ictal spikes, but which appear to be spiking independently of the channels involved in the serial sequences. Interestingly, although these ‘independent’ spiking channels do not participate in inter-ictal spike sequences, they do elicit large spike triggered potentials in many non-ictal channels. We also show that regions which participate in spike sequences are more likely to be functionally related, suggesting that it may be possible to identify the pathways which propagate inter-ictal spikes from functional connectivity maps. Conclusions: We show that there are temporal sequences of inter-ictal spikes which are overrepresented given the assumption of independent spike generation. Regions which participate in serial inter-ictal spike sequences are more likely to be functionally related, suggesting that inter-ictal spikes may be propagated through existing functional pathways which can be identified using functional connectivity measures. Funding: This work was supported by the Intramural Research Program at the National Institutes of Health.