Expert System Design for Evaluation of Muscular Atrophy Pritish Ranjan Pal, Abhas Kushwaha, Niladri Prasad Mohanty, Bikesh Singh, Bidyut Mazumdar Department of Biomedical Engg. National Institute of Technology, Raipur, Chhattisgarh, India pritishpal@ymail.com Abstract: Electromyography (EMG) signal is electrical manifestation of neuromuscular activation due to which physiological processes are accessible which cause the muscle to generate force and produce movement and help us to interact with the world. EMGs have large variation and nonstationary properties. There are two issues in the classification of EMG signals. One is the feature selection, and the other is classifier design. In EMG’s diseases recognition, the first and the most important step is feature extraction. In this paper, thirteen features have been used to classify EMG signals of isometric contraction for two different abnormalities namely ALS (Amyotrophic Lateral Sclerosis) which is coming under Neuropathy and Myopathy and the classification approach is termed as Muscular Atrophy Diagnostic Approach (MADA) by authors. For the issue of classifier design we apply Support Vector Machine classifier to classify between the signals as said above. From the experimental results, waveform length is the best feature with accuracy of around 94 percentage comparing with the other features, whereas features like Root mean square, spectrogram, kurtosis, entropy and power are other useful augmenting features. Keywords: ALS, isometric contraction, kurtosis, MADA, Myopathy, support vector machine. I.INTRODUCTION Human muscle consists of large number of fibers functionally arranged into individual motor unit which are all activated by nerve impulse from the nervous system which propagate through the length of the nerve fiber [1][2]. Electrodes placed at the region of these muscles pick up different voltages in different region of these impulses. A plot of these voltages is called as Electromyogram (EMG).These signals have the properties of nonstationary, nonlinear, complexity, and large variation [1]. EMG can be recorded by two types of electrodes which are an invasive electrode called as wire or needle electrodes and a noninvasive electrode called as surface electrode [3]. Use of lotions and creams on skins for 24 hour is avoided prior to EMG recording. EMG is a main nerve diagnostic tool used in Electro diagnostics medicine and in clinical neurophysiology. EMG test collect information about injured nerves, damaged nerves and muscle disorders leading to symptoms like numbness stinging, burning pain and weakness. EMG test is a valuable diagnostics tool which provides a real map for physicians which will help to evaluate locate and treat neuromuscular disorders [4]. Muscle diseases are mainly categorized by their clinical appearance. At the beginning of 1990‟s long lasting classifications of disease were based on the genetic abnormalities muscle and further it is also classified on the basis molecular abnormalities. Myopathy and Amyotrophic lateral sclerosis (ALS) are the two main muscular disorders and 20,000 Americans have ALS, and an estimated 5,000 people in the United States are diagnosed with the disease each year. Myopathy is diseases of skeletal muscle which are not the cause of nervous disorder [6] [7]. These diseases cause the skeletal or voluntary muscle to become weak or weakest. Myopathy is usually degenerative, but they are sometime caused by drug side effects, chemical poisonings, or a chronic disorder of the immune system. Amyotrophic lateral sclerosis (ALS), sometimes called Lou Gehrig's disease, is a rapidly progressive, invariably fatal neurological disease that attacks the nerve cells (neurons) responsible for controlling voluntary muscles characterized by the gradual degeneration and death of motor neurons [5]. About 5 to 10 percent of all ALS cases are inherited. Electromyogram is an important test for the detection of myopathy and ALS. In this paper, we propose a new technique, Muscular Atrophy Diagnostic Approach(MADA) which is based on feature extraction from acquired EMG for muscular diseases classification. EMG classification is one of the most difficult pattern recognition problems because there usually exists small but numerous variations in EMG features, which leads to difficulty in analyzing EMG signals. In muscles diseases recognition, there are two main points, namely feature selection and classifier design. In general, the methods of feature selection can be divided into two types: the measure of classification accuracy and the valuation using statistical criterion. After that the selection of the best features based on the proposed statistical criterion method is investigated. Proceedings of Second International Conference on Signals, Systems & Automation (ICSSA-11) 24-25 January, EC Department, G H Patel College of Engineering & Technology, Gujarat, India ISBN: 978-1-6123-3002-0 158