Over the last decade, several methods for analysis of epileptiform signals in electroencephalography (EEG) have been proposed. These methods mainly use EEG signal features in either the time or the frequency domain to separate regular, interictal, and ictal brain activity. The aim of this work was to evaluate the feasibility of using functional connectivity (FC) based feature extraction methods for the analysis of epileptiform discharges in EEG signals. These signals were obtained from EEG-fMRI sessions of 10 patients with mesial temporal lobe epilepsy (MTLE) with unilateral hippocampal atrophy. The connectivity functions investigated were motif synchronization, imaginary coherence, and magnitude squared coherence in the alpha, beta, and gamma bands of the EEG. EEG signals were sectioned into 1-s epochs and classified according to (using neurologist markers): activity far from interictal epileptiform discharges (IED), activity immediately before an IED and, finally, mid-IED activity. Connectivity matrices for each epoch for each FC function were built, and graph theory was used to obtain the following metrics: strength, cluster coefficient, betweenness centrality, eigenvector centrality (both local and global), and global efficiency. The statistical distributions of these metrics were compared among the three classes, using ANOVA, for each FC function. We found significant differences in all global (p < 0.001) and local (p < 0.00002) graph metrics of the far class compared with before and mid for motif synchronization on the beta band; local betweenness centrality also pointed to a degree of lateralization on the frontotemporal structures. This analysis demonstrates the potential of FC measures, computed using motif synchronization, for the characterization of epileptiform activity of MTLE patients. This methodology may be helpful in the analysis of EEG-fMRI data applied to epileptic foci localization. Nonetheless, the methods must be tested with a larger sample and with other epileptic phenotypes.