GaborPDNet: Gabor Transformation and Deep Neural Network for Parkinson’s Disease Detection Using EEG Signals

• DOI: 10.3390/electronics10141740
• Publication Date: 2021-07-20T09:10:59Z
• Authors: Hui Wen Loh; Chui Ping Ooi; Elizabeth Palmer; Prabal Datta Barua; Sengul Dogan; Turker Tuncer; Mehmet Baygin; U. Rajendra Acharya
• Keywords: electroencephalogram (EEG); 610; deep learning; Gabor transform; Classification; Parkinson’s disease (PD); 03 medical and health sciences; 0302 clinical medicine; classification; 616; spectrograms; CNN

Parkinson’s disease (PD) is globally the most common neurodegenerative movement disorder. It characterized by a loss of dopaminergic neurons in substantia nigra brain. However, current methods to diagnose PD on basis clinical features Parkinsonism may lead misdiagnoses. Hence, noninvasive such as electroencephalographic (EEG) recordings patients can be an alternative biomarker. In this study, deep-learning model proposed for automated diagnosis. EEG 16 healthy controls and 15 were used analysis. Using Gabor transform, converted into spectrograms, which train two-dimensional convolutional neural network (2D-CNN) model. As result, achieved high classification accuracy 99.46% (±0.73) 3-class (healthy controls, with without medication) using tenfold cross-validation. This indicates potential simultaneously automatically detect their medication status. The ready validated larger database before implementation computer-aided diagnostic (CAD) tool clinical-decision support.