Neuromodulatory Control of Bionic Limbs Enables Natural Walking Patterns
Post by Shireen Parimoo
The takeaway
Prosthetic or bionic limbs have traditionally required pre-programmed gait controllers to enable movement following limb amputation, but their use has not quite resembled natural walking gait. Now, researchers have shown that biomimetic gait (i.e., a gait that resembles that of biologically intact limbs) can be restored by providing neuromodulatory input to bionic limb controllers in real time through an innovative surgical procedure.
What's the science?
Bionic limbs have been around in some form for thousands of years and have undergone incredible advances in recent decades. Today, bionic limbs use controllers pre-programmed with gait algorithms to assist individuals with locomotion. However, they do not use sensory neural inputs to adjust gait, nor do they provide enough information about the position and velocity of the bionic limb to the user, often causing them to rely on visual inputs to control their movement. As a result, bionic limbs have not been able to fully mimic the natural gait of biologically intact limbs.
One reason why neurofeedback has not been fully integrated into bionic limb function is that the peripheral tissue that’s important for neuromuscular control is removed during the amputation surgery. The agonist-antagonist myoneural interface (AMI) is a procedure that uses the remaining biological tissue by connecting agonist (shortening) and antagonist (lengthening) muscles in the amputated region, to replicate the interaction between agonist and antagonist muscles present in biologically intact limbs. This week in Nature Medicine, Song and colleagues used AMIs to achieve biomimetic gait in individuals with bionic limbs without relying on pre-programmed gait controllers.
How did they do it?
Participants underwent the below-knee AMI amputation or a non-AMI amputation (i.e., control group). During the AMI surgery, the lateral gastrocnemius and tibialis anterior were connected to facilitate ankle control while the peroneus longus and tibialis posterior were linked to enable foot rotation. The AMI participants were fitted with a bionic ankle with a controller that monitored the angular position and speed of the ankle and an electromyography sensor that provided neurofeedback to the controller. Inputs from both the sensor and the bionic limb were processed by the controller for bidirectional feedback.
The authors measured walking speed and gait on a flat surface, on a slope (incline and decline), up and down stairs, and over small obstacles on the ground. Gait was assessed by examining the angular position and velocity (i.e., state) of the bionic ankle, while the torque angle (twisting or rotating) of the bionic ankle was used to quantify the degree of biomimetic gait. The ankle state and torque angle were used to calculate the amount of power and net work (total force) generated by the bionic ankle across different terrains. These measures were compared between the AMI and control groups, as well as with publicly available data on the gait patterns of individuals with biologically intact limbs.
What did they find?
The AMI group walked faster and showed a higher degree of biomimetic gait than the control group. For example, ankle power tends to increase with walking speed in biologically intact limbs, and this increase was seen only in the AMI group. The maximum walking speed of participants in the AMI group was also on par with that of individuals with biologically intact limbs. When participants were divided according to the magnitude of the agonist-antagonist muscle afferents, those in the control group had little to no residual muscle afferents while those in the AMI group had ‘moderate to high’ levels of afferents. Conversely, the torque angle trajectory of control participants resembled stiff ankles and non-biomimetic gait, while the AMI group were closer to exhibiting natural walking dynamics. Thus, using the residual muscle afferents to provide neurofeedback to bionic limbs enables biomimetic gait.
The AMI group adapted to different terrains faster than the control group. Their ankle peak power and net work increased while walking up an incline, while their shock absorption increased while descending stairs. The control group, on the other hand, did not show these adaptations and had a more limited range of motion. In the presence of obstacles, the AMI group increased dorsiflexion during the upswing of the ankle, followed by increasing propulsive power and net work on the recovery step, which allowed them to better maintain their walking speed. Dorsiflexion did not increase in the control group (and even decreased for some participants) in the presence of obstacles, which in turn slowed them down. Together, these results indicate that the neuromodulatory mechanism in the AMI amputation allowed participants to successfully adapt to different types of terrain and environmental obstructions.
What's the impact?
This study is the first to use AMI to demonstrate biomimetic gait and natural walking patterns in a bionic limb that relies entirely on neuromodulatory mechanisms. These findings show promise for AMI amputation to change the landscape of neuroprosthetic limb design and push the field closer to fully restoring biomimetic motor functioning following surgical amputation.