Brain region localization: a rapid Parkinson's disease detection method based on EEG signals

Area of Science:

• Neuroscience
• Biomedical Engineering
• Artificial Intelligence

Background:

• Parkinson's disease (PD) diagnosis relies on motor symptoms, leading to potential misdiagnosis.
• Early detection and monitoring of PD progression are vital for patient outcomes.
• Electroencephalogram (EEG) signals offer a non-invasive approach for potential diagnostic biomarkers.

Purpose of the Study:

• To develop and validate a novel method for diagnosing Parkinson's disease using EEG signal analysis.
• To explore the efficacy of 3D time-frequency spectrograms and AI in PD detection.
• To establish EEG as a reliable biomarker for early PD diagnosis.

Main Methods:

• Utilized two public EEG datasets for analysis.
• Constructed 3D time-frequency spectrograms from distinct brain regions using Continuous Wavelet Transform (CWT).
• Encoded spectrograms in RGB color space and trained ResNet18 models, validated with Leave-One-Subject-Out Cross-Validation (LOSOCV).

Main Results:

• Achieved high classification accuracies of 92.86% and 90.32% on the two independent datasets.
• Demonstrated the effectiveness of analyzing regional EEG spectrograms for PD detection.
• Validated the potential of the proposed AI-driven approach for PD diagnosis.

Conclusions:

• The novel approach using EEG signal analysis shows significant potential as a diagnostic tool for Parkinson's disease.
• The method provides a promising avenue for early and accurate PD detection, aiding clinical decision-making.
• Further research can explore refining this technique for broader clinical application in neurodegenerative disease diagnosis.