Diagram of the e-dermis system. Illustration: Osborn et al., 2018/Science Robotics
To make it work, the researchers created a neuromorphic system—a device that mimics the behavior of the nervous system with circuits. In this case, their neuromorphic model took the output of the e-dermis (i.e. the tactile information produced when touching an object) and transformed it into electric spikes, or neural signals, that replicated the behavior of mechanoreceptors and nociceptors. These spikes were then used to electrically stimulate the peripheral nerves of an amputee volunteer (i.e. transcutaneous nerve stimulation, or TNS). When provided with this nerve stimulation, the volunteer was able to feel sensations in his artificial hand.
In experiments, an amputee volunteer could feel pressure, the tapping of a fingertip, and even objects that elicited painful sensations. He could tell the difference between non-painful and painful tactile perceptions, including variations in an object’s curvature and sharp edges. The volunteer said the sensations felt like they were coming directly from the so-called phantom hand. EEG scans taken during the experiments appeared to show that regions in the brain associated with the hand were activated in the participant’s brain.
The JHU researchers documented which stimulations the user found painful and which felt more like normal touch. The volunteer was asked to rate the discomfort of the perceived sensations in the phantom hand using a scale from -1 to 10, where -1 is something enjoyable or pleasant, 1 is very light pain like an itch, 2 is a discomforting feeling like a pinch, 3 is uncomfortable but tolerable, like an accidental cut, and so on, During this experiment, the highest level of pain was ranked as a 3.
The volunteer performing the sensation task. Image: Osborn et al., 2018/Science Robotics
“One of the most surprising aspects of this work was being able to identify different stimulation patterns that produce different sensations in the phantom hand of the amputee volunteer,” said Osborn. “In this case, those sensations were of pressure or pain.”
To make the system more life-like, the researchers also added an automated pain reflex to the system. When the prosthetic hand touched a sharp object, the fingers automatically jerked away, “to prevent damage and further pain,” as the researchers write in the study. Importantly, the volunteer had no control over these reflex movements.
Sharlene Flesher, Sharlene Flesher, PhD, a postdoctoral researcher at Stanford University who wasn’t involved with the new study, said the new study is “a good piece of work that’s very complete,” and that “the progression they present is solid.” That said, she felt the EEG results were “silly.”
“They claim that it demonstrates that the participant felt the sensations in the left hand, but EEG does not provide the spatial accuracy to claim that,” Flesher told Gizmodo. “The result agreed that the sensations were on the left side of the body and probably somewhere on the arm, I did not buy that it was in the hand from the EEG report. I would have liked to see more detail about how they mapped the sensations, but it appears that they did a good job finding stimulation sites that evoked sensations in the phantom hand.”
As for building prosthetics that allow users to feel pain, Flesher agrees that triggering full-on pain should not be the goal.
“Whether or not pain should be relayed is interesting, and they kind of get at it here. If the prosthesis can identify ‘painful’ situations and minimize them, does discomfort really need to be relayed to the user? I think if they keep the pain sensations in an informative range, where it doesn’t cause so much pain so as to be a distraction, it’s useful,” she said. “However, they also evoked sensations with different modalities, such as pressure and tingle, so one reasonable pain-free alternative would be to have the tingle sensation indicate a painful touch. That being said, if they can evoke pain, pressure and tingle, using all three could convey more information.”
This is very promising work, but there are many other aspects of touch. Looking ahead, the JHU researchers would like to explore other perceptions that could be provided through sensory feedback, including temperature and proprioception (such as knowing the relative location of our body parts, like an arm above the head).
“By adding in different sensations, we can continue to improve upper limb prostheses to make them even more functional and lifelike,” said Osborn.
[Science Robotics]