EMG signal processing and diagnostic of muscle diseases

Abstract

Real time recordings of motor unit action potential (MUAP) signals from myopathy (MYO), neuropathy (NEU), and normal (NOR) subjects, using intramuscular electromyography (needle EMG) are treated and processed in order to be classified for the diagnosis of neuromuscular pathology. Feedforward-backpropagation neural network is used for the classification. Recognition rates were found to be higher than 70% and higher when using time domain features as inputs for the neural network.