Early detection of Parkinson’s disease based on beta dynamic features and beta-gamma coupling from non-invasive resting state EEG: Influence of the eyes

Abstract

Resting state electroencephalography (EEG) has been shown to provide relevant information for detecting neuropathological changes of the brain’s electrical activity in neurodegenerative patients. Studies conducted on local field potential recordings have shown that exaggerated beta oscillations and abnormally high beta-gamma phase amplitude coupling (PAC) are hallmark Parkinson’s disease (PD) signatures. Extracting beta bursts from non-invasive magnetoencephalography has also been found to be feasible, as it provides a better signal-to-noise ratio than electroencephalography and is less affected by volume conduction.

It is still unclear whether beta burst dynamic features and beta-gamma PAC from resting state EEG can be used to assess the progress of PD. In the present study, it has been proposed to assess the potential utility of beta burst dynamic and the beta-gamma PAC to discriminate PD patients from healthy subjects, as well as their relationship with clinical symptoms. Resting state EEG data have been analysed in both eyes closed (EC) and open (EO) and reactivity-to-eyes opening (REO) of a public database consisting of 20 healthy and 13 Parkinson patients. Beta burst events from EEG spectrograms were extracted to determine their dynamic features, i.e. burst duration, rate, peak frequency, spectral bandwidth and power as well as the normalized beta-gamma PAC. Permutation test while controlling the family-wise error rate was used to assess statistical significance. The results indicate that REO is more sensitive than EC and EO alone, and also that the higher variability of burst duration is linked to PD, while the lower burst rate is negatively correlated with clinical symptoms. PD patients had a higher periodicity of duration in the left frontal area, and a higher periodicity of peak frequency, spectral bandwidth and power of the bursts in the left central area than healthy subjects, together with a significant positive correlation with clinical symptoms.

Beta-gamma PAC not only found abnormalities in the central regions but also in the frontal, fronto-central, parietal and occipital regions, suggesting impaired motor, working memory and visuospatial skills. It was also possible to extract beta burst dynamic features and the beta-gamma PAC from resting state EEG and that these provided reliable PD progress biomarkers. These advances are expected to help clinicians design patient-personalised therapies and improve their quality of life.