Publishing house Radiotekhnika

"Publishing house Radiotekhnika":
scientific and technical literature.
Books and journals of publishing houses: IPRZHR, RS-PRESS, SCIENCE-PRESS

Тел.: +7 (495) 625-9241


Research in artificial neural networks the role of temporal parameters of inhibition in the local synchronization of activity in the cortex


V. G. Marchenko, K. A. Saltykov

One manifestation of synchronization in the cerebral cortex is the slow oscillation (epileptiform discharges) the periodicity of 0.3–1 Hz. Low-frequency rhythmic activity was observed in the EEG in certain types of anesthesia, as well as in a state of natural sleep. Between slow oscillations during sleep and the discharges of "spike-wave" there is a certain kinship. Discharges "spike-wave" may gradually evolve from slow (< 1 Hz) oscillations during sleep. As shown by numerous studies on the isolated cortex, slow oscillations are developed not only in the intact brain, but also in isolation of the cortex from the thalamus. Thus, the slow epileptiform potentials generated under certain conditions, neural networks of the cortex via intracortical mechanisms. It is assumed that the most probable factor providing synchronization in cortical neuronal ensembles at the time of slow oscillations was long hyperpolarization. A very important factor for realizing our model is that the slow oscillations were characterized by rhythmic cycles of depolarization and the generation of action potentials (Up states) followed by membrane hyperpolarization and cessation of discharges (Down states). The aim of our study was to determine, using the simulation model of the neural network, which parameters of activity (excitation and inhibition) should have the neurons to provide periodic synchronous activity, similar to the epileptiform discharges in the real neural networks of the cortex. To create a model of neural networks used the program "Neuroimitator" version 4.2. The model was a matrix of 20-69 at the same level of formal neurons, not directly connected with each other, as well as neurons in chronically isolated area of the cortex after long-term experience. Input effects on the formal neuron model was adjusted so that their activity is the period between excitation and inhibition, and it responds to the real properties of pyramidal neurons in the cortex. For one group of neurons were established short duration of inhibition in the range 75 – 200 ms, and for the other – in the range of longer duration 300-750 ms. Each neuron of these parameters in the cycle of excitation – inhibition were their own, and they were repeated several times within 10 seconds of the model. The degree of synchronization in the model was determined as the percentage of simultaneously active neurons in a certain period of time. The contribution of each neuron in the field potential was adopted the same. It seems to us, the most significant result of our work is the fact that we have created artificial neural networks for a sufficiently long period of time managed to mimic epileptiform activity with characteristics that are similar to that observed in neurophysiological experiments. The most important parameter of the simulation model was the frequency of occurrence of the moments when both active (the synchronous) a significant portion of neurons. The model points with a maximum synchronization activity of neurons were observed with a periodicity 1.4 seconds, i.e. a frequency of 0.7 Hz. In our earlier data in experiments on rats with application of penicillin to the cortex epileptiform potentials occurred after 1.6 seconds, i.e. a frequency of 0.6 Hz. Thus, in our model studies have determined the parameters of the activity of neurons in which synchronous activity is generated at intervals as close to that observed in real neural networks. Neurons, which make the greatest contribution to the synchronous activity had a duration inhibitory pause in the range from 100 to 150 ms and 300 ms. As follows from experimental data GABAB receptors are associated with a slow hyperpolarization, in contrast to fast GABAA. In the early phase of the burst response develops GABAA component with a duration of hyperpolarization in an average of 50 ms. While GABAB – develops more slowly and was observed in the late phase response with a peak of activity after 200 ms. For network operation mode, when the number of simultaneously active neurons at approximately twice the equivalent noise (i.e. number of neurons with respect to the non-marked rhythm), you must have a network of at least 75% of the neurons with the duration of inhibitory pause of 100 to 150 ms and 300 ms. Synchronization of neural activity in local neuronal networks that were studied in this paper was produced by a certain time the impulsed activity of neurons in the complete absence of connections between them. Spike activity of neurons was, in turn, was structured through the inhibitory time windows of different durations. Based on the results obtained on the model, we can assume that the synchronization of epileptiform discharges in local networks of the cortex is provided by inhibition of inhibitory pause lasting from 100 to 150 ms and 300 ms, associated with GABAB receptors. This assumption is supported by the data in the literature on termination absence epilepsy GABAB antagonists in animals. The activity of other receptors, as GABAA, and GABAB with other durations of hyperpolarization, in this case, for various reasons, poorly expressed.

© Издательство «РАДИОТЕХНИКА», 2004-2017            Тел.: (495) 625-9241                   Designed by [SWAP]Studio