Abstracts

Rationale:
Our goal is to determine if there are any distinguishable patterns of change in phase oscillation patterns before and during the onset of an epileptic seizure, and to evaluate the spatial scale over which these phenomena occur. Electrophysiological biomarkers associated with implanted electrodes have not been closely examined on a very fine spatial scale. Sudden phase oscillations detectable by time-frequency analysis techniques are related to cortical phase transitions, which likely change in frequency and spatial distribution as an epileptic seizure develops.    
Method:
We studied a 34 years old woman with refractory complex partial seizures who underwent standard intracranial subdural grid recordings to establish the localization of seizure onsets. In addition, after receiving IRB approval and informed consent, a 1x1 cm, 8x8 contact, subdural microgrid with 1.25 mm interelectrode separation was applied over right inferior temporal gyrus, near the seizure-onset zone.  A 100 sec section of micro-ECoG data obtained before and during a seizure was selected and analyzed. Data was imported into MATLAB, collected at 420 Hz, down-sampled to 200 Hz and filtered in the 1-50 Hz band. Hilbert transform was applied to compute the analytic phase, unwrapped and de-trended to disclose sudden phase changes. Phase oscillation rates were computed with a stepping window of one sec duration and step size of 5 msec. The phase oscillation rate (radians/sec) was computed for theta (3-7 Hz) and alpha (7-12 Hz) bands.  Spatiotemporal contour plots of phase change rates were constructed using a montage layout of 8x8 electrode positions, with phase change rates at 5 msec intervals. In a separate analysis using the same preprocessing steps as described, phase in the bands of interest was estimated using complex Morse wavelet convolution. We produced a video representing the evolution of phase change across the microgrid, with a grid of circular graphs representing the absolute value of the instantaneous phase of each sample of the seizure in each grid electrode. A raw timeseries waveform signal from a single electrode annotated with a moving star marking the location in the signal corresponding to a given frame of phase animation.  
Results:
The mean of the phase oscillation rates in theta band before and during the seizure were 45±7 and 22±3, respectively.  For alpha band these rates were 13±5 and 7±3, respectively. Spatiotemporal plots exhibited dynamic, oscillatory formation of phase cone-like structures in theta and alpha band frequencies. Phase cone structures were of greater amplitude before the seizure and progressively decreased in amplitude as the seizure developed. In contrast, the micro-ECoG amplitude was higher just prior to and during the seizure. The spatial phase animation demonstrated evolving phase architecture across the microgrid from the preictal state, progressing to population-level phase synchrony during the seizure.          
Conclusion:
These preliminary results suggest that sudden phase oscillations and spatial phase cone formations related to cortical electrical activity change before and during an epileptic seizure. Methods of instantaneous phase estimation resolves phase architecture on finely spaced electrodes. These findings may represent potential markers, suggesting that small brain regions may demonstrate complex spatial arrangements of phase synchrony both before and during a seizure, yielding unique insights in characterizing cortical areas involved in the initiation and propagation of ictal activity.  
Funding:
:None