Spatio-temporal sEMG image enhancement and motor unit action potential (MUAP) detection: algorithms and their analysis

In spatiotemporal multi-channel surface electromyogram (EMG) images where the x-axis is time, the y-axis is EMG channels and the gray level is EMG amplitude, the motor unit action potential (MUAP) appears as a linear Gaussian structure. The appearance of this MUAP pattern in the spatiotemporal images is mostly distorted either by the destructive superposition of other MUAPs occurring in the conducting volume or by various noises such as a power line, bad electrode and skin contacts and movement artifacts. For accurate automatic detection of MUAP, EMG image enhancement is needed to suppress the background noises and enhance the line-like MUAP propagation patterns. This study presents several candidate filters to enhance the MUAPs propagation pattern in spatiotemporal EMG images. The filters, which can detect and enhance line-like structure in digital images, are used. Specifically, the Hermite shape filter is used for EMG image enhancement and compared with Gabor filter and steerable filters. The performance of the filters regarding accuracy, specificity, and sensitivity is evaluated with real sEMG signal measured from different muscles and computer-generated EMG signals. In the enhanced images the visibility of the MUAP region is improved. These results can help in better estimation of muscle characteristics from sEMG signals.