Leg amputees are often not satisfied with their prosthesis, even though the sophisticated prostheses are becoming available.
One important reason for this is that they perceive the weight of the prosthesis as too high, despite the fact that prosthetic legs are usually less than half the weight of a natural limb.
Researchers led by Stanisa Raspopovic, a professor at the Department of Health Sciences and Technology, have now been able to show that connecting the prostheses to the nervous system helps amputees to perceive the prosthesis weight as lower, which is beneficial for their acceptance.
Together with an international consortium, Raspopovic has developed in recent years prostheses that provide feedback to the wearer’s nervous system.
This is done via electrodes implanted in the thigh, which are connected to the leg nerves present there. Information from tactile sensors under the sole of the prosthetic foot and from angle sensors in the electronic prosthetic knee joint are converted into pulses of current and passed in to the nerves.
“To trick an above-knee amputee’s brain into the belief that the prosthetic leg was similar to his own leg, we artificially restored the lost sensory feedback,” says ETH professor Raspopovic.
In a study published last year, he and his team showed that wearers of such neurofeedback prostheses can move more safely and with less effort.
In a further study, the scientists were now able to show that neurofeedback also reduces the perceived weight of the prosthesis. They published the results in the journal Current Biology.
In order to determine how heavy a transfemoral amputee perceives their prosthetic leg to be, they had a voluntary study participant complete gait exercises with either neurofeedback switched on or off.
They weighed down the healthy foot with additional weights and asked the study participant to rate how heavy he felt the two legs were in relation to each other. Neurofeedback was found to reduce the perceived weight of the prosthesis by 23 percent, or almost 500 grams.
The scientists also confirmed a beneficial involvement of the brain by a motor-cognitive task, during which the volunteer had to spell backwards five-letter words while walking. The sensory feedback not only allowed him to have a faster gait but also to have a higher spelling accuracy.
“Neurofeedback not only enables faster and safer walking and positively influences weight perception,” says Raspopovic. “Our results also suggest that, quite fundamentally, it can take the experience of patients with an artificial device closer to that with a natural limb.”
Recent advances in intelligent, powered prosthetic legs have opened up opportunities for individuals with lower limb amputations to restore their normative walking patterns on uneven terrains [1–9]. These modern devices use primarily autonomous control, which is, however, inadequate to assist other important daily tasks that involve unpredictable, non-cyclic motor behavior and require continuous coordination with the user’s motor control and environments. One example of such activities is anticipatory and compensatory postural control in standing, walking, or other recreation activities [10, 11].
Focusing on standing postural control, lower limb amputees wearing passive prostheses have shown decreased postural stability and increased compensation from the intact limb [12, 13]. This is partly because of the lack of active degrees of freedom in the prostheses. Powered prostheses have active, controllable joints and, therefore, a potential to enhance the amputees’ postural stability.
Unfortunately, there has been no autonomous control solutions to assist amputees’ standing posture because it is difficult to predict the postural perturbations and human motor control strategy for counteracting the perturbations. We are aware of only one research group, developing autonomous prosthesis control to assist posture stability of the prosthesis users when standing on slops .
The controller detected the inclination angle after the prosthesis foot was on a slope and then adjusted equilibrium position of prosthesis joint in order to support the amputee’s standing posture. This automated control was reactive and limited in function because it can assist standing posture on a slope only, and it acted only after the prosthesis foot was on an incline.
Hence, this prosthesis control was insufficient to assist anticipatory postural control (i.e., action before the perturbation happens) or handle the postural control under dynamic perturbations (e.g. weight transfer), which requires continuous postural control based on the shift of center of mass (COM).
As human neural control system is highly adaptable to the task context, perhaps neural control of prosthetic joint can be a viable solution to assist the amputee’s postural control and balance stability. EMG signals of the residual muscles are readily-available efferent neural sources in amputees and has been used for neural control of prosthetic legs in walking. Many groups have used EMG pattern recognition to classify the user’s locomotor tasks, switching autonomous prosthesis control mode accordingly for enabling seamless locomotor task transitions [15–19].
Another group used EMG signal magnitude recorded from the residual medial gastrocnemius (GAS) to proportionally modulate a control parameter in the automated prosthesis control in the push-off phase . Both aforementioned approaches relied on autonomous prosthesis control laws and cannot produce neural control of prosthetic joints continuously.
Three other groups conducted case studies to show the feasibility of direct EMG (dEMG) control in walking, in which EMG magnitude of one or a pair of residual antagonistic muscles are directly mapped to modulate the applied torque to the prosthetic joints continuously [21–23].
In dEMG control, the behavior of prosthetic joints is primarily and continuously determined by human neuromuscular control, mimicking human biological joint control. Note that the existing studies on EMG control of powered prosthetic legs, regardless the methods used, focuses on locomotor tasks mainly. Little effort has been aimed to address postural control.
One of the main challenges in implementing direct EMG control of a prosthesis, although technically simple, is whether amputees are capable of producing coordinated activations between the residual antagonist muscles for operating a prosthesis joint in dynamic task performance.
Previous studies have shown a large variation among transtibial amputees in producing coordinated activity between the residual tibialis anterior (TA) and GAS in a sitting posture or walking [22, 24, 25]. These results implied that individuals with transtibial amputations might no longer manifest normative activation in the residual muscles due to the limb amputation. Luckily, evidences have also shown that training or practice is a potential way to improve the capability of amputees in modulating residual muscles’ activity for dEMG control.
Our previous study  tested transtibial amputees in dEMG control of a virtual inverted pendulum, mimicking the dynamics of standing posture. We noted improved task performance for all the amputee participants after a short-term practice within the same experimental visit. However, the amount of improvement varied significantly among the participants. Acclimation to dEMG control has involved repeating the evaluated task (walking) for an extend period of time [21, 27], or visualizing phantom limb movements .
For Huang et al.  transtibial amputees did not adapt activation of their residual GAS until they were given visual feedback of their prosthetic ankle angle with a target trajectory. However, it was unclear whether, after removing biofeedback training, amputees could still reproduce desired ankle joint trajectories or continue to improve control. Dawley et al.  observed improvements residual muscle activity after simply repeated walking sessions with a single above-knee amputee.
From the findings of previous studies we postulate that amputees might adapt and learn the necessary muscle activation pattern for control function after training and practice. We expand the work of previous studies by
1) Creating and implementing an four-week PT-guided training paradigm without supplementary feedback of the ankle prosthesis
2) Implementing activity from both TA and GAS residual muscles for direct, continuous prosthetic ankle control and
3) Investigating the ability for an amputee to improve standing postural control with this control paradigm.
In this paper we present a case study to demonstrate the feasibility and potential benefit of dEMG control of a powered ankle prosthesis on an individual with a transtibial amputation for enhanced postural stability. Since training is likely necessary for successful application of dEMG control, the case study included four-week of physical therapist (PT)-guided training. Via this case study, we are interested in learning
(1) how a transtibial amputee learns residual muscle activation patterns and coordination necessary for driving a powered ankle prosthesis for postural control, and
(2) whether dEMG control of a powered ankle can improve the postural stability of transtibial amputees, compared to the currently prescribed passive prostheses.
The results may inform the needed training protocol for dEMG control of prosthetic ankle and the future development of versatile powered prostheses that can assist various activities of individuals with transtibial amputations.
reference link : https://www.biorxiv.org/content/10.1101/2020.09.11.293373v1.full
Source: ETH Zurich