INTRODUCTION

Modern prostheses have made strident gains in recent years, incorporating electromechanical components that are capable of mimicking human movement to rehabilitate, and restore function for persons living with musculoskeletal impairments. However, effective control of these devices has been hindered by limitations of current human-machine interfaces using amplitude-based myoelectric control schemes prone to variability and delay [1,2]. To address this healthcare need, we developed a human-machine interface, referred to as Motor Unit Drive (MU Drive), to provide control signals that are based on the firing behavior of individual motor units. Motor unit firing rates and recruitment provide natural physiological mechanisms for controlling force and movement in the intact limb [3] and therefore hold promise for more natural control of prostheses.

METHODS

We designed MU Drive algorithms using a series of processing stages to translate surface electromyographic signals recorded from single, dry, noninvasive, high-fidelity sensors placed over the muscles of a limb into measures of motor unit action potentials and firing times. The first stage of processing was developed to (Figure 1A) by verify sensor placement and acquire a-priori measurements of motor unit action potentials (MUAPs) from an initial brief contraction. The detected MUAPs are then used to resolve motor unit firings in real-time which are subsequently translated into biomechanically informed control signals using a neuromuscular transfer function between motor unit firing instances and physiologically derived estimates of motor unit force twitches (Figure 1B) to provide a control source that represents the intended movement. We evaluated the MU Drive signal characteristics and compared them to conventional amplitude-based myoelectric signals (Root-Mean-Square and Mean-Absolute-Value) recorded from healthy subjects (n=10) and subjects with congenital or traumatic trans-radial limb-loss (n=13) during actual and intended finger flexion/extension and forearm pronation/supination, respectively (Figure 1C).

RESULTS

In our testing of 23 subjects, we successfully measured the motor unit firing behavior from total of 4,942 volitional and intended contractions. On average, the MU Drive real-time performance achieved 2.7 ± 1.3 ms (5th and 95th Percentiles = 0.7, 4.7 ms) processing time for each 20 ms signal segment, equivalent to a real-time ratio less than 0.25:1 and upper-limit total delay of approximately 25 ms.

When we compared the MU Drive signal measured from each muscle with that of the amplitude-based signals obtained from the same muscle with a comparable 25 ms delay (i.e., 25 ms window), we observed that MU Drive improves the characteristic smoothness and responsiveness over the myoelectric alternative (Figure 2). Specifically, the MU Drive signal measured from the forearm pronator in the subject with trans-radial amputation during intended forearm pronation/supination maintained a smoothness of -6.0 (Figure 2A, black), a 97.4% improvement relative to the -232.0 smoothness measured from the RMS signal (Figure 2A, gray). Similarly, in the control subject during finger extension/flexion, the MU Drive signal measured from the finger extensors had a smoothness of -6.5 (Figure 2B, black), a 97.6% improvement relative to the -275.5 smoothness measured from the RMS signal (Figure 2B, gray) and closer to the -6.6 smoothness measured from the changes in finger angle of the control subject’s intact limb (Figure 2B, black dashes). When comparing the MU Drive and RMS signals measured from the finger extensors with the changes in joint angle of the 2nd digit of a control subject, we found that MU Drive was able to better replicate changes in movement of the intact limb with a relatively small error of 8.0% (Figure 2B, black), substantially lower than the 57.8% error measured from the RMS signal (Figure 2B, gray).

The test data showed no systematic differences in smoothness amongst the individual subjects, we grouped the smoothness data across all amputee subjects and separately across all control subjects for each muscle tested. We analyzed the smoothness measured from the amplitude-based signals across a range of window lengths (25 to 500 ms) that are commonly used for myoelectric control and tested them as a function of the latency with respect to the MU Drive signal obtained with 25 ms delay. Across all four muscles the median smoothness of movement of MU Drive was significantly greater (p<0.001) than the best median smoothness of the amplitude-based signals for both subject groups. Further analysis demonstrated that that the smoothness of the amplitude based signals was directly related to the latency of the signals with MU Drive, indicating the best performing windows for both functions lagged behind MU Drive with median latencies ranging from 49 to 171, and in most cases lagged behind the actual movement of the limb which had a median latency with respect to MU Drive that ranged from 61 to 159 ms. These data indicate that the smoothness of the amplitude-based measures can be improved, but at the expense of additional latency that results from increasing the window size.

Similarly, we grouped error data across all control subjects for each muscle and movement. The median error of MU Drive was significantly lower (p<0.001) than the measured median error of amplitude-based signals. Importantly we observed the percent error of amplitude-based signals were inversely related to the latency of the response with respect to MU Drive. These data give clear evidence that MU Drive signals are more responsive and more closely replicate the kinematics of the intact limb.

DISCUSSION

These results establish that MU Drive provides noninvasive, real-time access to natural control mechanisms of the human nervous system with significantly improved, smoothness, responsiveness and more faithful replication of the actual or intended limb movement when compared to amplitude-based signals. In fact, our analysis demonstrated amplitude-based signals are limited by an inherent tradeoff: increased smoothness and better replication of actual limb movement requires larger filtering windows that cause increased delay and reduced responsiveness. MU Drive overcomes the trade-off between performance and latency by using direct measures of neural commands that provide a smooth, proportional representation of physiological changes in joint position without the need for computationally expensive filtering. These results persisted across both control and amputee subjects tested, regardless of disuse and changes to muscle morphology due to surgery or congenital formation.

The proof-of-concept demonstrated in this work portrays significant advantages of MU Drive over conventional myoelectric human-machine interfaces. The improved performance characteristics of MU Drive are supported by the use of high-fidelity noninvasive sensors and real-time recognition algorithms to provide noninvasive access to commands of the nervous system for improving control assistive and prosthetic devices.

IMPLICATIONS

This work provides a foundation for future development of motor-unit-based controllers that leverage multiple MU Drive signals across synergistic muscles to enable simultaneous multi-degree-of-freedom control. The technology holds promise for improving assistive device function by achieving advanced control that better reflects user intent. It may also provide a new alternative for advancing the control of exoskeletons, and other rehabilitation applications.