EMG signals in their raw form are very rich in information, but in order to be valuable in the process of myopathy’s diagnostic, they have first to be amplified then analog to digital converted and filtered. In this paper we investigate the possibilities to realize an EMG measurement system with low cost hardware components and improve the performance of the implementation by signal processing. The measurement system has a minimal design consisting of an amplifier interface conform to typical myopathy data bases and an Arduino-based acquisition system. When electrodes are placed on the skin in the right position, the developed device generates signal form and values, which are similar to those downloaded from EMG database. For data acquisition, signal analysis and classification, an intelligent LabView based data processing software was implemented, which uses also a clinical standard database for normal and myopathic EMG signals. For classification, significant signal features were selected, which are: the Shannon entropy, variance, mean absolute value and signal positive peak amplitude. The investigation shows, that these features are sufficient to discriminate between normal subjects and patients with myopathy. A PCA guided K-means clustering classifier was established and a classification accuracy of 91.7 %.
If the inline PDF is not rendering correctly, you can download the PDF file here.