Experimental Apparatus and Data Collection

Experimental data from eight healthy subjects (6 male, 2 female; between 19 to 35 years of age) was obtained. Each subject provided written informed consent. The protocols were approved by the Institutional Review Board at the University of North Carolina at Chapel Hill. All methods were performed in accordance with the relevant guidelines and regulations.

Since individual fingers are controlled by separate compartments of the EDC muscle²⁷, selecting the precise location of each compartment was required to record corresponding MU activities of each finger. We located the different compartments based on our earlier high-density EMG study, which provide the information of spatial activation of individual compartments²⁷, and identified the appropriate electrode location that would yield a high EMG amplitude for a given finger extension. Then a five-pin sensor array (Delsys Inc., Natick, MA) is secured on the skin surface above a compartment. The EMG recordings from one sensor were considered to originate from a single compartment, given that the sensor array tends to give very selective EMG recordings from nearby muscle fibers based on our concurrent sensor array and intramuscular recordings²⁸. Specifically, if the intramuscular wire electrodes were inserted outside of the sensor array area spanned by the 5 pins, or if the wire was inserted deep to the muscle, the surface and wire electrodes could not record common MUs. In addition, the obtained action potential duration is also relatively narrow compared with the ones from the highly selective intramuscular electrodes, indicating that only a limited number of superficial fibers were recorded. These findings provide the evidence that the sensor array was selective enough to only record a small portion of the muscle. Since the locations of the two compartments which control the ring and little fingers were not readily distinguishable from the skin surface based on our previous study²⁷, we combined the recordings of the ring and little fingers together. The rectangular black boxes in Fig. 1 show the locations of the three compartments (index, middle, and ring-little) of the EDC muscle. In our study, two electrode arrays were used to concurrently collect EMG signals on a pair of separate compartments at any given time. Therefore, the electrode arrays were removed and reattached to different compartments, in order to obtain concurrent MU activities of three possible combination pairs (index vs. middle, index vs. ring-little, and middle vs. ring-little).

Block diagram of the data acquisition and analysis. Part 1 presents the high level neural control scheme. Part 2 presents data collection and EMG decomposition process. Part 3 presents the general coherence analysis method. This block diagram uses an example pair of the index and middle finger compartments.

Subjects were seated in a straight-back chair with their upper arm comfortably resting on a table, their right shoulder abducted at approximately 30°, the elbow angle extended at approximately 120°, the wrist in neutral position with 0° flexion/extension, and the metacarpophalangeal joint extended at approximately 120°. During the experiment, subjects were instructed to produce isometric finger extensions against a vertical board using all of their four fingers concurrently. A single-differential bar electrode (Delsys Inc., Natick, MA) placed on the mid-belly of the EDC was used to estimate the overall level of EDC activations, and was remained on the same location throughout the testing session, and the RMS value calculated using a 0.8 s moving window was provided to the subject as a feedback for the desired muscle contraction levels. The skin above the EDC muscle was scrubbed with alcohol pads to improve the electrode contact with the skin. Two five-pin sensor arrays (Delsys Inc., Natick, MA) were placed on the different compartment locations as described in the previous section. Five 0.5 mm diameter cylindrical probes were located at the four corners and the center of a 5 × 5 mm square. The electrode array recorded four channels of single differential EMG signals (Fig. 1). The sEMG signals were amplified by a gain of 1000, with a bandwidth of 20 Hz to 500 Hz, and sampled at 20 kHz.

The experiment consisted of two blocks. In the first block, subjects performed maximum effort of finger extension for three seconds, and the resultant RMS value of the EMG from the bar electrode was recorded. The subjects were asked to repeat the maximum effort three times with 60 s rest between trials, and the largest RMS value was then used to estimate the maximum EDC activation. In the second block, five isometric finger extensions (using four finger concurrently) were performed by the subjects. The RMS trace of EMG as in the first block was used as a feedback for the subject to track a target at 50% of the maximum RMS. The EMG recordings from the two sensor arrays were instantaneously shown to the operator. If the operator found that the subject was using one muscle compartment to compensate another, the subject was asked to repeat the trial. Specifically, the subjects were instructed to slowly increase their effect to reach the target and were then asked to maintain their RMS value at the prescribed level for 12 s. A 60 s rest period was provided between each successive trial to avoid cumulative muscle fatigue. The second block was repeated three times, in which EMG signals were obtained from the three different muscle compartment combinations. Therefore, a total of 72 (3 repeated trials ×3 compartment combinations × 8 subjects) trials were collected in this project with two sensor recordings in each trial. All the data were obtained within a single session in the same day for each subject.