Classification Of Normal People And People With Parkinson's Disease Based On Speech Signals Using Simulation Of MATLAB

Woon, Chee Hong (2017) Classification Of Normal People And People With Parkinson's Disease Based On Speech Signals Using Simulation Of MATLAB. Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text (24 Pages) Classification Of Normal People And People With Parkinson's Disease Based On Speech Signals Using Simulation Of MATLAB.pdf - Submitted Version Download (170kB)

Abstract

Parkinson’s disease (PD) is due to the death of dopaminergic neuron in the human brain. It will normally affect the people around the age of 50 to 60. Progressive loss of these dopamine neuron causing a various kind of motor and non-motor deficits. The latest statistic shows that the incident that people are suffering from PD is showing an increasing trend where men are having the higher possibility of suffering from the disease compared to women. However, early stages of the disease are difficult to diagnose when symptoms are often misdiagnosed with other medical conditions. Though the disease is hard to diagnose in its early stages, early diagnosis of the disease is necessary since medications are most effective at an early stage of the disease. Speech impairment is one of the earliest indicators for PD; therefore early diagnosis of the disease can be performed by using the speech signals. The criteria for designing a system that can be used to detect PD are pre-processing through end-point detection, feature extraction techniques, speech classifiers, database, and performance evaluation. The speech data acquired from the recording of speech signals of 40 subjects (20 PD patients and 20 normal subjects). In this project, feature extraction algorithms used are Linear Predictive Cepstral Coefficient (LPCC), Mel Frequency Cepstral Coefficient (MFCC), and Wavelet Packet Transform (WPT). While in classification, two types of classifiers are used which are Support Vector Machine (SVM) and Probabilistic Neural Network (PNN). The analysis shows that SVM classifier performs well in LPCC features while PNN classifiers perform well in MFCC and WPT features.

Actions (login required)

View Item