Parkinson disease early detection using EEG channels cross-correlation

Background: Nowadays, Parkinson’s Disease (PD) is the second most common neurodegenerative disease worldwide and has no known cure. Researchers are focused on delaying the disease progression and, to this end, it is necessary to detect PD as early as possible. Methods: This work proposes a method to detect early stages of PD using the cross-correlation function between selected electroencephalogram (EEG) channels, which is used to find time delays at scalp level in order to serve as a metric for distinguish study groups. Genetic algorithms and a binary classifier (X-ROC) were used to preselect the best features combinations to feed a neural network for data classification. Results: This novel approach achieved an overall accuracy of 92.66%.Conclusion: The applied method detects relevant early PD activity at scalp level in right temporal, frontal, parie to-occipital and occipital regions with high accuracy.