Dynamic Analysis And Control Of Epileptiform Spikes In A Neural Mass Model

Electroencephalography signal (EEG) is the integrated performance of thepostsynaptic potentials for a large number of neurons which contains plenty of physiologyand disease information, what provides the important basis for brain disease study, such asepilepsy clinical diagnosis and treatment. The neural mass model can describe real EEGsignals more accurately. The research for model-based close-loop control can provides atheoretical basis for the actual epilepsy treatment.In this paper, the epileptiform spikes dynamic characteristic for the neural massmodel is analysed, and the different control schemes for suppressing the epileptiformspikes are researched. We do the following work:First, it provides the dynamic analysis of the epileptiform spikes for the neural massmodel. It is proved that the excitement may lead to the seizures through the sequencecharts which are gained by adjusting the excited parameter for a single neural mass model.For the three coupling neural mass model, the adjustment of the coupling parameterproves that seizure is probable association with the coupling factor. Second, a fuzzy PIDcontroller is designed to trace epileptiform spikes in a neural mass model and thesimulations prove the track effect. The effect of parameters on the control energy and theeffect of the types of controlled populations on the ability of realizing track are analyzed.Third, the adaptive unscented Kalman filter algorithm is proposed considering thelimitation that UKF can achieve excellent filtering performance only for the certain noise,and the control scheme is designed to suppress epileptiform spikes in a neural mass modelwith uncertain measurement noise. The difference is shown through comparing theadaptive unscented Kalman filter control to the regular unscented Kalman filter control.Finally, it makes the PCB to simulate the brain wave based on system on programmablechip (SOPC), and makes the neural mass model that can produce epileptiform spikessignal for one or three coupling neural mass model, what set a basis for the further studyof hardware controller.