Parkinsons disease detection based on image analysis of EEG signals

Abstract

Parkinsons disease (PD) is a common chronic neurological disease, that causes great disturbance to the patient's life and work, and when the disease develops seriously, it may even lead to the death of the patient. Until now, treating PD has been a tough nut to crack and a financial challenge for families and governments alike. In this paper, we propose to use the Resnet-50 Neural Network model to differentiate between 41 PD patients and 41 normal subjects by analyzing time-frequency domain maps of electroencephalography (EEG) signals. This method achieves classification accuracies ranging from 81% to 85% for six-channel detection and varying from 76% to 77% for single-channel detection, which opens up new avenues for the early diagnosis of Parkinson's disease, demonstrating the potential to combine EEG signals with image processing.