EEG-Centered Parkinson's Disease Prediction System Using Gaussian Kernel Discrete Wavelet Transform and Leaky Single Peak Triangle Context Convolutional Neural Network Deep Learning Technique

Abstract

Parkinson's disease (PD), a neurodegenerative illness where dopamine-producing brain cells are destroyed, slows down motor performance. To identify PD, numerous techniques have recently been established. In this research, a novel idiosyncratic Gaussian Kernel Discrete Wavelet Transform (GKDWT) and a Leaky Single Peak Triangle Context Convolutional Neural Network (LeakySPTC-CNN)-centered effective Deep Learning (DL) model have been developed for the prediction of PD. The Electroencephalogram (EEG) signal is first acquired and then divided into one-second segments. The model then carried out signal filtering and spectrogram conversion. The features are then extracted with the support of GKDWT, and feature selection and ranking are carried out using the Quadratic Chimp Optimization Algorithm (QChOA). The LeakySPTC-CNN is used to categorize both ill and healthy people. The model has a promising future, according to the experiential evaluation, and it also offers a better way to diagnose Parkinson's disease (PD) early.