The Identification Of Psychogencit Nonepileptic Seizures Based On Resting-state EEG

Epilepsy is a syndrome of recurrent and unprovoked seizure, with abnormal excessive neuronal activity in brain. The main diagnosis method is the expected electroencephalographical epileptic changes. Psychogenic nonepileptic seizure(PNES) is similar to epileptic seizure, but lacks the expected electroencephalographical epileptic changes. Video-EEG(vEEG) combined with medical history provided by patients and witness could be used as the diagnostic criteria of PNES. Even so, many PNES patients are still misdiagnosed to epilepsy, which would result in the delayed diagnosis and therapy. The established PNES early diagnostic approaches require detailed information, but the collection of those information is very time consumig. Even worse, there may be no vEEG in some places, and the ictal process may be hard to be recorded for some patients. If the effective alternative methology could be developed for early PNES diagnosis, it will be of helpful for clinical doctors.In recent years, the analysis based on graph theory has been widely used as a tool to study the anatomical and functional network of central nervous system. This disseration based on resting-state EEG is to extract the differentiating features between PNES and epilepsy from the perspective of network topology, aiming to provide the feasible and robust features for the related diagnosis.The main contributions of this disseration are as follows:1. Using theoretical analysis method to investigate the network properties of PNES, epilepsy and normal control group. The results show that the network properties of PNES and epilepsy group were significantly different from the normal group. However, there is no statistical network properties difference bewent PNES and epilepsy, which results in the difficulty to discriminate them based on the network properties.2. Developed the spatial pattern method to extract the differenting spatial topology of brain network existing between PNES and epilepsy. The conducted study reveals that 92.00% accuracy, 100.00% sensitivity, and 80.00% specificity were reached for the classification between PNES and focal epilepsy when the proposed approach is utilized.3. The above analysis procedures are integrated into the three isolated GUI modules accounting for the related processing including extraction of network properties, spatial pattern analysis, and the classification, respectively.The conducted study reveals that the functional brain network topology is of significant difference between PNES and epilepsy patients, but this abnormal difference cannot be well reflected by the statistical network properties. The conducted study also demonstrated that the developed spatial pattern analysis can actually capture this differentiating spatial topology information, and provides a helpful tool for PNES diagnosis to clinical doctors.