Please use this identifier to cite or link to this item: http://imsear.hellis.org/handle/123456789/129768
Title: Modeling an epileptic brain using discrete and continuous neural network models
Authors: SELVARAJAN, S
Keywords: Models, neurological
Issue Date: 2004
Publisher: University of Sri Jayawardenepura: USJ(MED).
Citation: SELVARAJAN, S, Modeling an epileptic brain using discrete and continuous neural network models, University of Sri Jayawardenepura USJ(MED), 2004: x,109p.
Abstract: Various neural network models on epileptic behavior of the human brain were investigated. Several experimental studies have been carried out to simulate behavior of human epileptic brain recently. One of such experiment is the study carried out by Schiff etal. To study the firing behavior of neural networks in hippocampus slices of rat brain. In that study human epileptic brain activity was introduced using the high potassium concentration ([K+]?), where slices from the hippocampus of the temporal lobe of rat brain where exposed to artificial cerebrospinal fluid. Before introducing the high potassium concentration, it was observed that signals recorded from brain slices contained no spikes while with the introduction of high potassium concentration, spikes appeared in random intervals as in the case of brain having epilepsy. The brain-slice experiment mentioned above has been examined using both discrete and continuous neural network models. This discrete model was based on the model developed by Biswal et.al. and Dasgupta et.al. (BD model). In this model, the effect of high potassium medium was introduced through a Hebbian learning mechanism which is switched on during the simulation under reduced inhibition. The sub-passes which play a crucial role in reproducing experimental results in BD model are found to be not necessary when random weights at different stages of the simulation are introduced. A continuous biophysical neural network model was also developed to describe the outcome of the brain-slice experiment mentioned above. In addition, effect of the input current and the potassium concentration changes on dynamics of a single neuron and population of neurons were investigated. It was found that the maps of network activities exhibits stable stationary states and bursting states like trajectories similar to those were found in experiments on hippocampus slices. The discrete and continuous neural network models developed in this wirk were able to successfully reproduced the experimental results.
Description: Dissertation: M.Phil., University of Sri Jayawardenepura: USJ(MED), 2004.
URI: http://imsear.hellis.org/handle/123456789/129768
Appears in Collections:University of Sri Jayawardenepura

Files in This Item:
There are no files associated with this item.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.