Classification of myopathy, neuropathy and healthy groups from electromyography signals

Electromyography (EMG) is a technique used to analyse the electrical activities of muscles It is especially used to examine the physiological state of muscle tissue called striated muscle or skeletal muscle in the body. Two different electrodes can be used to obtain the EMG signal. One of them is surface electrodes and the other is an invasive electrode type called needle electrode. The most common medical use is the needle electrode. The EMG signal is the sum of the nerve action potentials that occur in each spindle of the muscles. This signal varies according to a healthy individual in patients with signs of myopathy (muscle disease) and neuropathy (nerve disease). In this study, a decision support system capable of scoring myopathy and neuropathy with time and frequency analysis of the clinical EMG data obtained from the EMGLAB database was attempted. Mean absolute value, zero crossing rate and willison amplitude analyzes were performed in time plane. Analyzes were made for power spectral density on the frequency plane. In addition, the EMG signal is separated into sub-bands by discrete wavelet transform method and the frequency analysis of the second approximate subband is to calculate the power spectral density by the welch method and various attributes are extracted from this power spectral density. Successful classification results were obtained with different machine learning methods trained using these attributes.