Epilepsy Analysis And Control Based On Neural Mass Model

Epilepsy is a chronic neurological disorder characterized by recurrent and paroxysmal epileptic seizures. Epileptic seizures result from abnormal, excessive or hypersynchronous neuronal activity in the brain, which manifest as high-amplitude, irregular excessive discharges in electroencephalogram (EEG). The pathogenesis and treatment of epilepsy have been major and hot issues in neurology. The lumped-parameter neural mass models (NMMs) with neurophysiologically based parameters can build a bridge between signal processing and neurophysiology, which can be used for studying the mechanisms underlying seizures and for controlling seizure. Based on the improved NMMs, this thesis conducts the following studies in the aspects of neurophysiological feature analysis of EEG during ictal and interictal periods, controlling epileptogenic excitation and synchronization.1) To study the neurophysiological mechanisms underlying EEG signals, EEG signals are regarded as the outputs of single NMM or multi-NMM respectively, then the model parameters for EEG signals during different periods are identified, and the differences of neurophysiological mechanisms among different periods of seizures are discussed by comparing the parameter distribution. A delay unit and a gain unit were added to the Wendling model to fit EEG signals in time domain, which simulates the amplifier and time delay in the process of recording EEG depending on EEG device. Parameter identification of improved Wendling model can been regarded as an optimization problem, genetic algorithm (GA) is used to identify an optimal set including of five parameters to minimize the error between real and simulated EEG signals. The model parameters denoting the strength of inhibition and excitation were identified for two sets of EEG signals recoding during ictal or interictal periods respectively by the presented method. The results show that the model with identified parameters can simulate the real EEG signal well. Through twenty times trials for every selected EEG signal, it is found that the dispersion of the identified parameters is small in most cases. This means that the identification procedure with GA is stable. By comparing the identified parameters for ictal and interictal EEG signals, the differences of excitation and inhibition between ictal and interictal EEG signals were discussed. It is less reasonable that EEG signals are all regarded as the outputs of single NMM with identical model parameters. The outputs of NMM are simple because the diversity of neural in a same NMM is ignored. To fit real EEG signals better, a multi-NMM is proposed. The multi-NMM includes multiple NMMs, and its output is the linear combination of the outputs of all NMMs. Then EEG signals are regarded as the outputs of multi-NMM, the NMM number of which is unfixed, and is minimized under the premise of guaranteeing fitting effect. The problem above can be simplified as a constrained l0 norm minimization problem, orthogonal matching pursuit (OMP) is used to solve it. The results show that the NMM number during ictal period is significantly less than during interictal period, and the strength of major NMMs increases apparently. This shows that massive neural fuse into larger neural masses and produce high-amplitude discharges. Moreover, it is found that the distribution of excitatory and inhibitory strength during ictal and interictal periods is similar, but the excitation/inhibition ratio during ictal period is higher than during interictal period.2) Over-excitatory neuronal activity has always been considered as the leading cause of epileptic seizures, abnormal internal feedback which can not keep the balance between excitation and inhibition may trigger seizures. To regulate the seizures caused by over-excitation, two control policies were presented. One is decreasing excitatory strength, the other is increasing inhibitory strength. Epileptiform index is presented to denote the seizure degree and used as control variable of proportional-integral-derivative (PID) controller to control epilepsy seizures. Neural mass model (NMM) is used as a test-bed to simulate the change of seizure degree with the increase of excitatory strength and two control policies. Experimental results show that the increase of excitatory strength can lead to a substantial increase of epileptiform index and trigger seizures. PID controller which is used to decrease excitatory strength or increase inhibitory strength can keep excitation-inhibition balance and inhibit epileptic seizures.3) Epileptic seizures are always accompanied by hypersynchronous firing of neurons, which manifests as high-amplitude, irregular excessive waveform in electroencephalogram. In an effort to understand the synchronization mechanisms underlying epileptic seizures and how to avoid them, a model composed of several neural masses was built. Principal component analysis (PCA) was used to identify synchronization clusters composed of several neural masses. A method for calculating the synchronization cluster strength and participation rate is presented. The synchronization cluster strength can be used to identify synchronization clusters and the participation rate can be employed to identify neural masses that participate in the clusters. Each synchronization cluster is controlled as a whole using a PID controller without epileptogenic zone localization. We illustrate these points using coupled neural mass models of synchronization to show their responses to increased (between node) coupling with and without control. Experiment results indicated that PID control can effectively regulate synchronization between neural masses and has the potential for seizure prevention.4) Exact localization of the epileptogenic zone (EZ) is the first priority for ensuring epilepsy treatments and reducing side effects. The results of traditional visual methods for localizing the origin of seizures are far from satisfactory in some cases. Signal processing methods could extract substantial information that may complement visual inspection of EEG signals. In this study, EZ localization is changed into a driver identification problem, and a nonlinear interdependence measure, the weighted rank interdependence, is proposed and used as a driver indicator because it can detect coupling information, especially directionality, from EEG signals. A proportional integral derivative (PID) controller is then explored, using simulations, to establish its suitability for seizure control. The seizure control we propose rests on identifying the EZ using nonlinear interdependence measures of directed functional connectivity. Two directionally coupled neural mass models are employed for simulation investigation. Two parameters can adjust the sensitivity and completeness of the weighted rank interdependence for different applications, and their effect is discussed in the context of neural mass models. Simulation results demonstrate that use of the weighted rank interdependence for EZ identification can be applied to different EZ types, and the approach achieves an overall identification rate of 98.84% for several EZ types. Simulations also indicate that PID control can effectively regulate synchronization between neural masses.NMM can be used as an effective complement of clinical trials for research in neurophysiology, which has some advantages such as low cost, easily adjusting parameters, this study will help drive more application of NMM in different fields.