Applications of neural networks to medical signal processing

The authors describe two applications of NN (neural networks) that are intended to provide an illustration of some of the capabilities of NN-based medical signal processing. The first concerns the decomposition of surface (transcutaneous) EMG (electromyography) into its inaccessible motor unit action potentials (MUAP) for the purpose of determining the sequence of single (or groups) of MUAP which form the corresponding surface EMG signal. The second application concerns the discrimination between intended walking functions of paraplegics, activated by functional neural stimulation (FNS), from parameters of signature patterns of above-lesion surface EMG. The first application is mainly of diagnostic significance, though it also has functional potential, as a possible approach to control FNS in paraplegics. The other application is mainly functional (for controlling FNS), though diagnostic uses are possible