The Effect of Vector Length and Gain Quantization Level on Medical Parameters of EMG Signals on Lossy Compression

In this work, we have used the shape-gain vector quantization, a lossy compression method, to compress data from EMG signals. The focus of the study was to reveal the effect of segment length in vector quantization and the number of gain quantization levels in preservation of the main clinically interesting medical parameters of the EMG signals: mean frequency and median frequency. The experimental results show that there is segment length dependent high band noise over 300 Hz and segment length dependent attenuation in spectra under 300 Hz. The effect of gain quantization upon the medical parameters is minor.