NEURAL-NETWORK ANALYSIS OF THE EMG INTERFERENCE PATTERN

This paper investigates the performance of artificial neural networks for analysing and classifying EMG signals from healthy subjects and pa tients with myopathic and neuropathic disorders. EMG interference patt erns (IF) were recorded under maximum voluntary contraction from the r ight biceps of a total of 50 subjects. Parameters were obtained from t he signals using recognized quantification techniques including turns analysis, snail segments analysis and frequency analysis. Supervised n etworks examined were an improved backpropagation network (IBPN), a ra dial basis network (RBN), and a learning vector quantization network ( LVQ). Supervised networks using different combinations of parameters f rom turns analysis and small segments analysis gave diagnostic yields of 60-80%. Combinations using frequency analysis parameters produced s imilar results. The performance of unsupervised Self-Organising Featur e Maps (SOFM) was generally lower than that of the supervised networks . Including personal data (sex and age) did not improve the overall pe rformance.