Brain Connectivity Measures in EEG-Based Biometry for Epilepsy Patients: A Pilot Study

The use of electroencephalography (EEG) signals for biometrics purposes has gained attention in the last few years, and many works have already shown that it is possible to identify a person based on features extracted from these signals. In this work we focus on four functional connectivity measures (magnitude-squared and imaginary coherence, motif synchronization and space-time recurrence) for the classification of 10 epilepsy patients with recorded resting-state EEG signals, to compare and discuss different methodologies. We perform the analysis by slicing the signals of at least 2 trials for each subject in epochs of 3 and 10 s, filtering the data in the ranges of 1 40Hz and 1 100 Hz, building reference and test vectors from the connectivity measures and labeling each test vector to a subject using the minimal Euclidean distance from the feature vectors. The best classification rates were obtained with magnitude-squared coherence and motif synchronization, for the data segmented in epochs of 10 s and filtered between 1 40Hz. All the measures with the signal filtered in the same range obtained an accuracy equal or higher than 80%, a result that can be enhanced with more complex classifiers.