How different skin preparation methods affect surface electromyographic measurements in the horse

Skin preparation is paramount for surface electromyographic (sEMG) signal quality and subsequent measurement results. However, no standardised skin preparation procedures for animals have been reported. In our study, four skin preparation procedures for measurements using sEMG in horses were compared and evaluated. Five muscles (longissimus dorsi, semitendinosus, triceps brachi caput longum, ulnaris lateralis and gluteus medius) of five warmblood horses were measured under four skin preparation conditions each. These conditions were: (1) no preparation (hair length ±5 mm); (2) cleaned (ethanol 80%); (3). clipped (1 mm) and cleaned; and (4) shaved and cleaned. Electromyographic data were collected (TMSi SAGA, 4,000 Hz, bipolar configuration, pregelled Ag/AgCl electrodes) for 60 seconds at trot (12.5 km/h, treadmill) for each condition. Input impedance, root mean square (RMS) and signal-to-noise ratio (SNR) were compared using linear mixed models (fixed effect=‘condition’, random effect=‘horse’ and covariate=‘muscle’). No contact was achieved (invalid input impedance values) in 25/25 ‘no preparation’ and 19/25 ‘cleaned’ conditions. Qualitatively, the remaining six ‘cleaned’ trials showed unrealistically high raw signal values and did not resemble equine muscle activation patterns. All trials for conditions 1 and 2 were thus removed from quantitative analysis. Differences in estimated marginal means between ‘clipped’ and ‘shaved’ were -0.4 kΩ (P=0.03), 0.1 μV, and 1.4 dB for input impedance, RMS and SNR, respectively. The results demonstrate that without hair removal, sEMG signal quality is very poor and does not resemble the muscle activation pattern in horses, which will lead to unreliable study outcomes. In addition, the differences between clipping and shaving, both combined with cleaning, are neglectable.