FUNCTIONAL CONNECTIVITY AND GRAPH-THEORETICAL ANALYSIS FOR NON-INVASIVE PRESURGICAL EVALUATION IN DRUG-RESISTANT FOCAL EPILEPSY

Epilepsy affects approximately 50 million people worldwide and is characterized by recurrent seizures. In patients with drug-resistant focal epilepsy (DRFE), surgery is considered an effective strategy for reducing seizure activity. We propose a presurgical evaluation framework integrating brain connectivity and graph-theoretical analyses of the pre-ictal to ictal transition to identify epileptogenic zones. A cohort of patients with DRFE who underwent brain magnetic resonance imaging (MRI) and long- term scalp electroencephalographic (EEG) recording was included. Cortical sources were reconstructed from scalp EEG by projecting electrodes onto MRI-based cortical surfaces. Temporal Granger causality analysis was applied both ten seconds prior to seizure onset (pre-ictal phase) and at seizure onset (ictal phase). Graph-theoretical centrality measures and hubness were used to estimate outcomes and assess lateralization across regions. Thirteen patients (7 females) were recruited. Seven underwent surgery, with six achieving favorable outcomes; functional connectivity analysis correctly identified both lateralization and localization in four patients, and lateralization only in one. One case had motion artifacts. In the patient with poor outcome, a bilateral epileptic network was suggested. Among four Stereo EEG-monitored patients, lateralization was correct in all, with concordant localization in one. Of the three non-operated patients, analysis indicated contralateral lateralization in one, unclear lateralization in another, and correct lateralization in the remaining case (Figure 1). Pinpointing epileptogenic zones through presurgical evaluation may be fundamental for clinical decision-making. By investigating seizure lateralization and epileptogenic foci, this approach may integrate analytical tools with visual inspection of EEG, optimizing presurgical evaluation and supporting surgical planning.