Research Library

A prosthetic limb can restore some functionality after an amputation, and muscles remnant in the residual limb are often used to generate signals to control it. However, in high amputation levels, such as above-elbow, there are not enough muscles left to control all the many missing joints. In this study, we demonstrated that splitting the nerves severed by the amputation and rerouting them into remnant and free muscles grafts can increases the number of potential control signals. This surgical approach, in combination with our neuromusculoskeletal interface, allowed an individual with above-elbow amputation to control all five fingers of a prosthetic hand intuitively.

Electrical stimulation of the nerves is known to elicit distinct sensations perceived in distal parts of the body. The stimulation is typically modulated in current with charge-balanced rectangular shapes that, although easily generated by stimulators available on the market, are not able to cover the entire range of somatosensory experiences from daily life. In this regard, we have investigated the effect of electrical neurostimulation with four non-rectangular waveforms in an experiment involving 11 healthy able-bodied subjects. Weiss curves were estimated and rheobase and chronaxie values were obtained showing increases in stimulation time required to elicit sensations for some waveforms. The localization of the sensations reported in the hand also appeared to differ between waveforms, although the total area did not vary significantly. Finally, the possibility of distinguishing different charge- and amplitude-matched stimuli was demonstrated through a two-alternative-forced-choice (2AFC) match-to-sample task, showing the ability of participants to successfully distinguish between waveforms with similar electrical characteristics but different shapes and charge transfer rates. This study provides evidence that, by using different waveforms to stimulate nerves, it is possible to affect not only the required charge to elicit sensations but also the sensation quality and its localization.

Replacing human hand function with prostheses goes far beyond only recreating muscle movement with feedforward motor control. Natural sensory feedback is pivotal for fine dexterous control and finding both engineering and surgical solutions to replace this complex biological function is imperative to achieve prosthetic hand function that matches the human hand. This review outlines the nature of the problems underlying sensory restitution, the engineering methods that attempt to address this deficit and the surgical techniques that have been developed to integrate advanced neural interfaces with biological systems. Currently, there is no single solution to restore sensory feedback. Rather, encouraging animal models and early human studies have demonstrated that some elements of sensation can be restored to improve prosthetic control. However, these techniques are limited to highly specialized institutions and much further work is required to reproduce the results achieved, with the goal of increasing availability of advanced closed loop prostheses that allow sensory feedback to inform more precise feedforward control movements and increase functionality.

Decoding human motor intentions by processing electrophysiological signals is a crucial, yet unsolved, challenge for the development of effective upper limb prostheses. Pattern recognition of continuous myoelectric (EMG) signals represents the state-of-art for multi-DoF prosthesis control. However, this approach relies on the unreliable assumption that repeatable muscular contractions produce repeatable patterns of steady-state EMGs. Here, we propose an approach for decoding wrist and hand movements by processing the signals associated with the onset of contraction (transient EMG). Specifically, we extend the concept of a transient EMG controller for the control of both wrist and hand, and tested it online. We assessed it with one transradial amputee and 15 non-amputees via the Target Achievement Control test. Non-amputees successfully completed 95% of the trials with a median completion time of 17 seconds, showing a significant learning trend (p < 0.001). The transradial amputee completed about the 80% of the trials with a median completion time of 26 seconds. Although the performance proved comparablewith earlier studies, the long completion times suggest that the current controller is not yet clinically viable. However, taken collectively, our outcomes reinforce earlier hypothesis that the transientEMGcould represent a viable alternative to steady-state pattern recognition approaches.

Prosthetic embodiment is a complex phenomenon capturing how strongly a prosthesis is perceived to be part of our own body (ownership) and to what degree we are in control of the prosthesis (agency). We proposed a multi-dimensional framework to showcase that prosthetic embodiment depends heavily on which tasks a prosthesis is used for (degree of interaction with the environment) and which prosthetic technology is used to do so (more advanced prostheses offer more functionality and thus the basis for volition and multisensory input to be correctly integrated more frequently). This work provides the context required to better understand prosthetic embodiment research and thereby make its results more relevant and readily applicable to prosthetics outside the research laboratories.

When we electrically stimulate the nerves of people with amputation, they can feel sensations on their missing limb, which we can use to provide them with feedback from their prosthetic hand. However, these electrical stimulations may also be picked up on the implanted sensors used to control the prosthesis. When this happens, these stimulations can sometimes cause the prosthesis to move uncontrollably or unpredictably. To address this issue, we developed and tested two stimulation artifact removal algorithms can learn to filter out these stimulations from the prosthesis control signals. In doing so, we demonstrated the ability to improve the quality of these control signals, as well as to improve the user’s ability to control their prosthetic hand simultaneously with nerve stimulation. Overall, this will improve our people’s experience when using sensory feedback at home to feel the objects they are interacting with, without having to sacrifice their control over their prostheses.

We developed a new method to calculate the strength of electrical connections between sensors implanted in the body, such as those used with our neuromusculoskeletal prosthesis. This method relies only on simple mathematics, and can work for any type and number of implanted sensors that share an electrical reference. Using this method, we are able to paint a more complete and accurate picture of how well our implanted sensors are working, as well as to more quickly pinpoint the cause of issues when they arise.

Robotic prostheses controlled by myoelectric signals can restore limited but important hand function in individuals with upper limb amputation. The lack of individual finger control highlights the yet insurmountable gap to fully replacing a biological hand. Implanted electrodes around severed nerves have been used to elicit sensations perceived as arising from the missing limb, but using such extra-neural electrodes to record motor signals that allow for the decoding of phantom movements has remained elusive. Here, we showed the feasibility of using signals from non-penetrating neural electrodes to decode intrinsic hand and finger movements in individuals with above-elbow amputations. We found that information recorded with extra-neural electrodes alone was enough to decode phantom hand and individual finger movements, and as expected, the addition of myoelectric signals reduced classification errors both in offline and in real-time decoding.

Working under the assumption that steady-state EMG signals are more prone to be affected by factors such as differing physiology and fatigue, we investigate the possibility of creating a classifier that acts on the transient EMG observed during muscle contraction. During evaluation we obtained performances comparable to state-of-the-art steady-state EMG controllers; illustrating the possibility of using the transient portion of EMG signals for classification.

When one user of our neuromusculoskeletal prostheses competed in the Cybathlon 2020, an international competition for advanced bionic limbs, he not only trained intensely during the months leading up to the competition, but also continued using his prosthesis more often and dexterously even after the competition had ended. By looking at data saved to a memory card in the prosthesis, we discovered that he learned to control his hand with more finesse, and the skills he gained from training carried over into his daily life as he uses his prosthetic hand, wrist, and elbow at home. When asking him about his experience, he told us “I don’t feel like I have one arm anymore,” and explained that he felt his self-confidence had improved from participating in the competiton. We hope that these results will encourage clinicians to use competitions like the Cybathlon to encourage their patients to train with their prosthetic limbs and to regain more of their lost function in their daily lives.