A novel signal processing method using system identification for underwater surface electromyography

Currently, to record underwater surface electromyography (EMG), electrodes are covered with waterproof tape. For short-term measurement, waterproof tape prevents electrical leakage. However, during long-term measurement, water or sweat can contact the electrodes, changing the measurement conditions and gradually affecting the EMG data. The purpose of present study was to devise a novel method for prolonged underwater EMG recording, which estimate dry-land EMG from underwater EMG recorded by non-waterproofed electrodes using system identification techniques.

One healthy male participated in this study. System identification was used to convert underwater EMG signals to the estimated dry-land signals. Transfer functions were derived using two pairs of surface recording electrodes on the same muscle in parallel. System input was the EMG recorded using non-waterproofed electrodes; the output was the signal recorded underwater using waterproofed electrodes (supposed to be the same as dry-land signals). To examine the validity of the present method, three experiments were conducted.

There was a high positive correlation between the estimated dry-land EMG based on the non-waterproofed electrodes and the EMG obtained using waterproofed electrodes. To test the validity of long-term recording using the novel method, the estimated dry-land EMG signals were measured during 30 minutes of underwater stepping and were stable.

The novel method using non-waterproofed electrodes with system identification techniques eliminated the effect of changes in measurement conditions and appears effective for long-term, underwater surface EMG recording.