Artifacts removal and application of cross-frequency coupling to EEG signals of epileptic patients

Epilepsy is a neurological condition characterized by recurrent episodes, called convulsive seizures, consisting of sudden and repeated muscle movements, often associated with temporary suspension of the state of consciousness. Seizures are generally of very short duration, although they can sometimes last for a long time, and are the result of an abnormal and prolonged synchronization of the activity of neurons in the cerebral cortex or brainstem. Through the use of an innovative approach, such as the association of EEG and functional magnetic resonance imaging, EEG-fMRI, it is possible to map the specific functional activation of certain brain areas. In particular, this thesis focuses on the study of the interaction between oscillations at different frequency bands. This is called cross-frequency coupling, CFC. The CFC analyses the mechanisms that regulate neuronal processing distributed between the various brain frequencies. This integration of the processing between frequencies could be obtained through the CFC, specifically the phase-amplitude coupling, PAC. Before studying CFC in our data, it was essential to proceed with the cleaning of signals from artifacts. Indeed artifacts, such as eye movements, blinking, heart signals, muscle noise, and power line interference present serious problems for the EEG interpretation and analysis. To remove artifacts from EEG recordings, especially those resulting from eye movements and blinking, many methods have been developed. In particular, in this thesis, the analysis of independent components for the removal of artifacts has been analysed and applied. Once the data has been cleaned, cross-frequency coupling was investigated before and after eye closing in patients with eyelids myoclonia with absences, a rare epilepsy syndrome with seizures evocated by eye closure plus seizures that appear spontaneously. Typically, seizures occur within 3 seconds after eye closure. The data were also collected in a rest condition, i.e. the situation where a subject does not perform any specific task.