Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)

Graduate Group


First Advisor

Brian Litt


Currently, clinical electrode arrays with a sparse spatial density (1 cm) are used to map the seizure onset zone (SOZ) and epileptic network in patients prior to epilepsy surgery. However, recent research demonstrates that submillimeter, cortical-column-scale domains have a role in seizure generation that may be clinically significant. We used novel high-resolution (500 μm), active, flexible, 1-cm2, surface, electrode grid-arrays to explore the behavior of these domains. We employed this new technology to analyze the two-dimensional (2-D), wave-propagation patterns of epileptiform local field potential spikes (LFP spikes). Subdural micro-electrocorticographic (μECoG) signals were recorded in vivo from anesthetized cats. A GABA antagonist, picrotoxin, was applied to induce acute neocortical epileptiform activity leading up to discrete seizures. Nine hours of data yielding 26,331 LFP spikes was analyzed. Features characteristic of spatio-temporal (ST) patterns were extracted from these events and k-medians clustering was employed to separate the data into 10 distinct classes. We tested the hypothesis that 2-D spike patterns during seizures (ictal spikes) are different from those between seizures (interictal spikes). A permutation test (n=1,000,000) confirmed this hypothesis. A frequent episode discovery algorithm (Temporal Data Mining) was then applied to investigate the relationship of sequences of these patterns to seizure generation, progression and termination. We found that sub-millimeter-scale ST spike wave-propagation patterns reveal network dynamics that may elucidate mechanisms underlying local circuit activity generating seizures. We conclude that sequences of patterns of similar type are less likely to precede seizure generation than sequences of patterns of differing types. Temporal analysis of these patterns also suggests that seizures in this model are not initiated by a single 2-D pathway, but rather by a number of different ST-initiating events. While these findings may be model-specific, reflecting diffusion of picrotoxin across the feline neocortex, the tools we have developed to interpret these events are directly portable to the human condition. We are confident that recording LFP spike ST wave-propagation patterns at high resolution provides a fruitful direction for continued analysis of epileptiform network dynamics and we propose that further study may provide a novel opportunity for therapeutic intervention at the micro-scale to treat epilepsy.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Included in

Biomedical Commons