Closed-Loop Control Of Epileptic Seizures: Model Analysis

Epilepsy poses a grave threat to people’s physical health, whose pathopoiesia and treatment mechanism is not clear. The electrical stimulation is an effective substitution of medicaments and surgery. Although closed-loop control is effective in treatment, it is rare using neuron and neural network to achieve closed-loop control. Substituting model analysis for animal and human body is an effective way. In this thesis, we combine computational neuron model with control theory and propose a closed-loop control strategy based on neuronal model, which achieve the closed-loop control of epilepsy.In this thesis we analyze the current situation and development of the research and control of epilepsy model, propose closed-loop control that use key parameters and slow variables as feedback parameters for CA3 area in the hippocampus. Through the dynamic characteristics of PR neurons analysis, the regularity that various parameters influence CA3 firing (that is the epileptics pathogenesis in the hippocampus) are obtained. We can find that various parameters in CA3 area have different time scale and slow variables also have great physiological significance. In this thesis, the closed-loop feedback control scheme which adopts slow variable tracking is proposed for the first time. Two slow variables of PR neurons (namely the intracellular calcium ion concentration and open probability of long time calcium-activated potassium channel) are chosen as feedback variables respectively. We choose Unscented Kalman Filter (UKF) to estimate the slow variables, and use the PI control algorithm to realize the control changing from rapid firing to slow firing and from firing to rest. We derive from the simulation that the control signal has fluctuations with smaller amplitude when using closed-loop control with slow variables as feedback parameters, which accords with physiological conditions.Considering that morbid firing of neurons is usually resulted by abnormalities of one or multiple parameters and these abnormalities can recover under external electrical stimulations, closed-loop feedback control based on key parameters are proposed for the first time. We estimate these key parameters in real-time using UKF and take them as feedback parameters to realize the closed-loop control and the tracking of normal parameters. Through simulation we found this scheme can adjust the abnormal firing caused by one parameter or multiple parameters.At last, an effect model under external electrical fields of single neuron and neuronal populations is built according to the actual physical environment with neurons embedded in the neural medium. Analyzing its non-linear dynamic characteristics, we obtain the key parameters of neurons and network under electrical fields and estimate those parameters using UKF, and then choose them as feedback signals to realize network closed-loop control. The simulation results show that our control scheme can realize the control of the desynchronization and the synchronization of network.In this thesis, the closed-loop control strategy of neurons and neuron populations under the external electrical stimulation is proposed and the dynamic characteristics of the system are obtained through model analysis. The simulation results demonstrate the effectiveness of the proposed scheme, and provide theoretical guidance for electrical stimulation therapy of neurologic diseases.