Abstracts

RATIONALE: Nonlinear mutual cross prediction (MCP) characterizes dynamic interdependence among nonlinear systems. MCP also reveal relative strength of thecoupling between systems, thus provides information about the direction of interdependence. The state of response system can be predicted by knowing the state of driving system alone. We analyzed MCP of EEG of two medically intractable temporal lobe epilepsy patients, who underwent left temporal lobectomy. METHODS: Scalp EEG was recorded interictal and ictal period during presurgical evaluation. Asymmetry of mutual prediction among all channels was investigated and then classified into driving and response system. Five epochs of interictal EEG free from epileptiform discharge(s) and that of a complex partial seizure of each patient were analyzed. Ictal period was subdivided to 3 parts, preictal, ictal and postictal and contaminated in parts with movement artifacts during seizure. These data were mapped in colors according to intensity. RESULTS: In interictal period, left frontal and both occipital region showed driving system, and right temporo-parietal region was seem as response system. Before onset of the seizure, the intensity of driving system was stronger in left frontal region. At the onset of seizure, driving system located on left frontal and anterior temporal region and then moved to mid-temporal region. In the middle of the seizure, the distribution of driving system was diffused throughout in left hemisphere with moderate intensity. Thereafter left anterior and mid-temporal was of strong driving force. Response system was located in right temporo-parietal preictally, centroparietotemporal region ictally, and right posterior half of brain postictally. The spatio-temporal distribution and propagation of ictal EEG was consistent with the results of visual analysis. CONCLUSIONS: MCP analysis may be a useful method for detecting interdependence of complex system. It reflects temporo-spatial distribution and the pattern of propagation in temporal lobe epilepsy, even for the EEG contaminated with movement artifacts.