Abstracts

Rationale: Neural network simulations using a model with realistic cortical architecture have been used to study synchronized bursting dynamics as a seizure representation (1,2). This model has the interesting property that bursting epochs arise spontaneously and cease at random intervals, making the onset of the seizure-like episodes difficult to predict. We have used this neural network simulation to study the time-frequency properties of this evolving bursting activity. Methods: The model represents a region of cortex of dimension 1.6 mm X 1.6 mm, and includes seven neuron classes organized by cortical layer, inhibitory or excitatory properties, and electrophysiological characteristics. There are a total of 65, 536 modeled single compartment neurons that operate on a modified version of Hodgkin-Huxley dynamics. The intercellular wiring is based on our previous modeling studies and is patterned after histological data. A spatially distributed Poisson noise source representing activity of neighboring cortex affects 1% of the modeled neurons. Time-frequency analysis of the spontaneous bursting patterns used the method of matching pursuits. The simulations are performed on a 16-node distributed 32-bit processor system. Results: The bursting phase is characterized by a flat frequency spectrum. Varying the random order of connections and the random sequence of background inputs produces similar stochastic behavior in the model. Increasing the frequency of noise application demonstrates a transition in the model to a regular high frequency spiking activity. Additionally, transitions from intermittent spiking behavior into constant bursting activity can also be demonstrated as the number of connections between major excitatory components is increased. Results from interburst interval histogram parametrization and Poincaré return maps are also presented demonstrating the stochastic nature of the bursting epochs (Figure 1). Conclusions: This large scale multi-neuron network simulation demonstrates random onset and cessation of the epileptiform behavior making the bursting transitions difficult to predict.In this model, seizures can occur infrequently, and they can be triggered by normal background activity originating from outside the epileptogenic zone. Detectable preictal changes here are not present; the first network changes being in fact seizure initiation (3,4). 1. Anderson WS, et al. Biol Cybern 2007;97:173-194. 2. Anderson WS, et al. Epi Res 2009;84:42-55. 3. Lopes da Silva F, et al. Epilepsia 2003;44(Suppl. 12), 72-83. 4. Mormann F, et al. Brain 2007;130, 314-333. PK supported by NIH NS051382.