Abstracts

RATIONALE: Determinations of the localization of neocortical epileptic foci are limited by the spatial resolution of existing methods. High temporal resolution methods are needed to differentiate ictal onsets from areas of early regional spread to facilitate localization in patients undergoing presurgical evaluations. We developed a computational model of epileptic foci in simulated neural networks with the goal of modeling difficult to localize clinically relevant events associated with seizure origination from areas with rapid regional spread.
METHODS: Our model incorporates: 1) an array of excitatory neurons capable of reproducing the spread of bursting activity and 2) a small sub-population or sub-network which triggers the activity in the network array. Neurons are modeled using a conductance-based model and realistic synaptic connections are simulated. Connections between neurons in the array are local e.g. each neuron receives inputs from neighboring neurons only. This network is capable of reproducing the spread of bursting activity. To simulate the epileptic focus, an additional sub-population of neurons is incorporated into the network array. This sub-network is comprised of neurons capable of generating endogenous bursts (model 1) or has the built-in structures of connections capable of maintaining self- sustained oscillations (model 2).
RESULTS: In both types of models of the epileptic focus we observed recurrent waves of bursting activity spreading through the network array. In the first model when intrinsically bursting neurons stimulate with the same period and phase postsynaptic neurons in the array we observed strictly periodic waves of bursting activity spreading through the network array. The frequency of propagating waves is determined by the recovery time of neurons from afterhyperpolarization and by the frequency of burst occurrence in intrinsically bursting neurons in the triggering subpopulation. When the number of intrinsically bursting neurons involved in triggering of the activity increases (e.g. the area of projection increases), the spatiotemporal pattern of activity is recurrent. In instances of activation of the network array by an [dsquote]epileptic circuit[dsquote] (model 2) single pulse stimulation of one neuron in the triggering network produces persistent activity in this network, which drives the postsynaptic neurons in the network array. The number of active postsynaptic neurons in the area of the projection of the focal ictal activity changes in time. In this type of focus (model 2) induced spatiotemporal patterns of activity in the network array are sensitive to oscillations in the area of projection from the focus.
CONCLUSIONS: The presented model can clearly reveal differences in the pattern of activity in the triggering zone (model epileptic focus) and in the surrounding network. This model reproduces synchronous bursting events in network arrays when the population of neurons in the focus (participating in the origination of ictal events) is relatively constant (model 1) or varies from moment to moment (model 2). These models can be helpful in investigations of the nature of ictal or non-ictal events near the region of ictal onset.
[Supported by: NIH grant NS38958]