Automatic detection of PTZ-induced seizures based on functional brain connectivity network in rats

Functional brain connectivity (FBC) network has been used for characterizing the dynamics of seizure evolution. In this paper, FBC networks based on correlation coefficient (COR), cross-correlation (xCOR), coherence (COH), phase slope index (PSI), phase locking value (PLV), phase lag index (PLI), mutual information (MI), transfer entropy (TE), Granger-causality index (GCI), directed transfer function (DTF), and partial directed coherence (PDC) were constructed. Graph theoretic measures, including transitivity, modularity, characteristic path length, radius, diameter, and global efficiency were calculated for each FBC network, and used as the feature vector for classifying ECoG signals corresponding to seizure and nonseizure. The results show that an average accuracy of 95.3% was obtained using the combined features from all FBC networks.