CLASSIFICATION OF T\TEUROMUSCIILAR DISORDERSI BASED ON ELECTROMYOGRAPTTY (EMG) SIGNALS

Electromyography (EMG) signals are the measune of activity in the muscles. The
motion of the muscles will be generated and recorded using skin surface electrodes.
EMG signals can be found from anywhere on the exterior of human's body such as
biceps, triceps, shoulder, arm, hand, leg. The aim of this project is to identifr the
neuromuscular diseases based on EMG signals by means of classification. The
ncuromuscular diseases that have been identified are healthy, myopathy and
neuropathy. The signals weIE taken and analyzed from EMG lab database to become
datasets for classification system. The classification was carried out using Artilicial
Neural Network. In this project, there are two techniques that used to classifr three
different types of muscular disorders such as Multilayer Perceptron (MLP) and
Wavelet Neural Network (WlIhI). And the input that applied to these systems using
feature extraction from EMC signals. In time domain, five feature extraction
techniques that used to exmct the sample of signal such as Autoregressive (AR),
Root mean square (RMS), Zero crossing (zc), waveform length (wL) and Mean
Absolute Value (MA$. The comparison between different techniques will be
included based on the accuracy of the result. The input data has been used in
Multilayer Perceptron (MtP) to train the classification system. Besides that,
frequency domain was used for extracting the useful information from EMG signal
for Wavelet neural network (wl[Nr) such as Power Spectrum Density (pSD), both
systems were hained and the test performances were examined after training to
provide the best result.