Abstracts

Rationale: Synaptic potentials produce traveling waves that spread to multiple brain regions. These waves can be observed in different cognitive states and with a wide range of speeds, 0.3-0.8 m/s for unmyelinated and 1-10 m/s for myelinated axonal fibers (Muller. L., 2018). They are characterized during seizures by high-amplitude low-frequency discharges with different delays across arrays of implanted electrodes. Originating from the Epileptogenic Zone (EZ). Previous studies have mostly focused on ictal discharges in field models, microelectrode recordings, and 2-dimensional grid structure electrodes. There are few studies that systematically analyze discharges as traveling waves in the 3-dimensional brain. The propagation networks of these traveling waves can potentially help us to differentiate between the EZ and penumbra, an adjacent area that potentially need not be resected or ablated. Here, we describe a new algorithm to construct these 3D networks based on intracerebral Stereoelectroencephalography (SEEG) recordings. Methods: Four patients who underwent SEEG implantation were selected. Criteria for selection were being seizure free after surgery and having a clearly defined and spatially limited EZ. We selected one seizure for each subject. Ictal data in the bipolar montage was bandpass filtered (2-40 Hz) and the early part of seizure recruitment was dropped (roughly 15s each subject). Using clinical findings and visual inspections of discharges delays in channels near the EZ, we defined a presumptive source (reference) for traveling waves. This was either a single contact pair for some patients or the average of two for others. We explored a propagation volume up to a 30mm Euclidian distance with respect to the source. A 4s sliding window with 3s overlap was used to compute the cross-correlations between the reference and propagation area. The temporal delay was defined as the shift resulting in the highest cross-correlation. The correlation values and time delays were computed for designated electrodes and time points. The corresponding p-values were calculated by comparing correlations with a null distribution extracted using shuffled data. We retained delay values with significant correlation (alpha = 0.05) after FDR correction. Finally, the median of surviving delay values was calculated for each channel. Results: Fig. 1. depicts the propagation pattern of ictal discharges in four patients in the 3D volume showing median propagation delay vs. Euclidean distance from the source. These results show a clear and significant correlation of delay with Euclidean distance. In these subjects, we can observe two traces of ictal discharge spreading in a manner consistent with macroscopic (fast) and mesoscopic (slow) wave propagation (Muller. L., 2018). Additionally, one can find several instances in which propagation occurs along the electrode characterized by increasing delays with respect to the sequential contacts on that electrode. As mentioned earlier, the seizure propagation network could be a useful tool to more precisely define the EZ.  For example, in a recent case study for the subject (M1988H3A) with unilateral periventricular nodular heterotopia (Cvetkovska. E., 2019), we observed that the V4-V5 channel displayed evidence of epileptogenicity. However, Fig. 1. shows a 40ms delay between this channel and the source, which might suggest ruling out this area as part of the EZ. This is consistent with the surgical strategy reported in (Cvetkovska. E., 2019) in which the V4-V5 area was not laser ablated and the patient was rendered seizure free. Conclusions: We presented our preliminary results for propagation networks in a three-dimensional volume. Results suggest that these mappings can be useful for distinguishing between the EZ and penumbra. Funding: This work was supported by the National Institutes of Health under awards R01NS089212, R01EB009048, and R01EB026299.