Localization Of Epileptogenic Zone Based On Stereo-electroencephalography

Epilepsy is a common and severe chronic neurological disorder characterized by recurrent epileptic seizures.For patients with refractory focal epilepsy,presurgical evaluation is required to identify the source and extent of the epileptogenic zone(EZ).Stereo-electroencephalography(SEEG)can record deep brain activities that cannot be observed with scalp EEG by implanting electrodes into brain regions stereotactically.In clinical,the EZ is currently identified by visually inspecting the ictal SEEG recordings,a procedure that is time consuming and inevitably affected by the drawback of subjectivity.In this thesis,a computional method based on SEEG and effective connectivity analysis was used to localize the EZ,and its effectiveness was also evaluated.In this thesis,the SEEG data were acquired from four patients who underwent presurgical evaluation of SEEG at the Department of Functional Neurosurgery in Renji Hospital.All patients were diagnosed as focal epilepsy and had good results after resective surgery.First,time variant multivariate autoregressive(TVMVAR)model was constructed for data 120 s before and 90 s after the ictal onset marked by clinician.The model was estimated by Kalman filter,whose parameter was determined through convergence evaluation.Then spectrum weighted time variant partial directed coherence(TVPDC)was computed to get dynamic directed network of the epiletic brain.After that,three graph measures,out-degree,in-degree and betweenness centrality,were used to analyze the characteristic of the dynamic network.Finally,the results 40 s before and 5 s after the ictal onset were averaged,and the positions of electrode contacts with relatively higher value were considered as indicative EZs.These indicative EZs were then compared to the clinically diagnosed EZs.In all four patients,the indicative EZs localized by in-degree or betweenness centrality were highly consistent to the clinically diagnosed EZs.However,the out-degree indicated that there was no significant difference between nodes in the network.In this work,the method based on ictal SEEG and effective connectivity analysis localized the EZ accurately,and could be applied readily in a clinical setting.It also suggested in-degree and betweenness centrality may be a better network characteristic to localize the EZ than out-degree.