Directed-transfer-function based analysis for epileptic prediction

Epilepsy is a neurological disease with the feature of repeated seizures. The probability to predict upcoming seizures is an issue that attracts both researchers and clinicians. Meanwhile, the directed transfer function (DTF) analysis as mechanism for feature extraction in the intracranial electroencephalographic (iEEG) recordings applied to epileptic prediction could reflect the dynamics changes of brain activity before and after seizure onsets. In the present paper, a cortical network calculated by the method of DTF was investigated with the iEEG recordings from a patient undergoing presurgical evaluation. Particularly, the closeness centrality metric was evaluated to identify critical network nodes in the cortical network during both interictal and preictal states of an epileptic seizure. Obtained values of the closeness centrality obviously increased at 20s and 40s preictal in onset 1 and onset 2, respectively. As a conclusion, a significant change of the closeness centrality during preictal states would be a useful indication to early predict an upcoming seizure in patients with epilepsy.