Tez No	İndirme	Tez Künye	Durumu
401367		Evaluating effects of denoising and feature extraction methods on classification of EMG signals / Yazar:ERCAN GÖKGÖZ Danışman: PROF. DR. ABDULHAMİT SUBAŞI Yer Bilgisi: International Burch University / Yurtdışı Enstitü Konu:Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrol = Computer Engineering and Computer Science and Control Dizin:	Onaylandı Doktora İngilizce 2014 179 s.
Machine Learning techniques and algorithms are extensively using in biomedical signal processing field to produce diagnosis system for definition of neuromuscular diseases. Different approaches have been applied for quantitative analysis of biomedical signals. Electromyography (EMG) biomedical signal data is used in this study to define neuromuscular diseases which consist of electrical currents from muscles during its contraction. Motor Unit Action Potentials (MUAPs) in EMG signals provides an important source of information for the diagnosis of neuromuscular disorders. This study shows how machine learning techniques are used for detection of neuromuscular diseases. EMG signals were recorded under usual and real conditions for quantitative analysis. The EMG signals were recorded from five places in the muscle at three levels of insertion. The dataset consist of control, a group of patients with Myopathy and Amyotrophic Lateral Sclerosis (ALS) data. EMG signals contain noise while traveling on different tissues. Moreover, the EMG signal acquisition collects signals from motor units at a time which may be effected by different signals. Numerous feature extraction and denoising methods applied to remove noise from signal and to extract features from signal. Feature extraction methods such as Autoregressive (AR) Burg, covariance (COV), multivariate covariance (MCOV), Eigen value, Multiple Signal Classification (MUSIC), discrete wavelet transform (DWT) and classifiers such as k-Nearest Neighbor (k-NN), Artificial neural network (ANN), Radial Basis Function (RBF), Support vector machine (SVM), Regression Trees (CART), C4.5 and Random Forest classifiers are utilized in this study. Multiscale Principal Component Analysis (MSPCA) denoising method applied to remove noise from EMG signal. The effect of the MSPCA denoising and feature extraction methods is discussed on EMG signal classification by applying different classifiers. The comparisons between the developed classifiers were based on a number of scalar performances such as sensitivity, specificity, accuracy, F-measure and area under ROC curve (AUC). The results show that MSPCA de-noising has considerably increased the accuracy as compared to EMG data without MSPCA de-noising. Combination of DWT feature extraction and Random Forest achieved best performance for k-fold cross validation with 96.67% total classification accuracy. These results demonstrate that the presented framework have the potential to achieve a reliable classification of EMG signals, and to support the clinicians for making an accurate diagnosis of neuromuscular disorders. Keywords: Electromyography (EMG), Motor unit action potentials (MUAPs), Multiscale Principle Component Analysis (MSPCA), ARburg (AR), Covariance (COV), Multivariate Covariance (MCOV), Eigen value, Multiple Signal Classification (MUSIC), discrete wavelet transform(DWT) and classifiers such as k-Nearest Neighbor (k-NN), Artificial neural network (ANN), Radial Basis Function (RBF), Support vector machine (SVM), Classification and Regression Trees (CART), C4.5 and Random Forest