Abstract
Amputee patients who have lost a hand or arm are severely impaired in their daily life, as they lose the ability to grasp and interact with their environment. While the use of electromyographically controlled prosthetic devices, such as robotic arms, do give back means to grasp objects again, making dexterous movements with them is still difficult and more importantly, they lack the ability to give sensory feedback. The sense of touch is not only critical for making simple movements as tying your shoelaces, but it also plays an important role in emotional communication and the embodiment of the limbs. Central nervous system interfaces do allow for bidirectional control of prosthetic devices, how- ever they are highly invasive and might give an abstract encoding of the subject’s intention. An alternative approach could be to extract movement information from the peripheral nervous system (PNS) instead. Beside the reduction of invasiveness, it could also greatly improve decoding, as PNS electrodes will record the direct feed to the muscles, and thereby could avoid the perhaps more complex signals of the CNS. Aside from that, stimulating the PNS, instead of the CNS could evoke more naturally perceived sensations of lost limbs. Even in forearm amputations the neural pathways are still preserved, potentially making PNS interfaces excel- lent candidates for bidirectional control of motor prosthetics. Recent development in electrode fabrication allows the production of very fine multichannel wire electrodes that can be inserted into the nerves. In this thesis, I investigated if a bidirectional prosthetic interface can be achieved using PNS ar- rays implanted in the medial and ulnar nerve in the upper limb of a rhesus macaque (Macaca mulatta). This was done done with Transverse Intrafascicular Multichannel Electrodes (TIMEs), which are 12 channel, thin-film electrode arrays capable of recording and stimulating individual nerve fascicles. Two acute and one short-term experiment with non-human primates (NHP) showed that it is feasible to implant TIMEs in the PNS of a rhesus macaque. With the surgical procedure established, two long-term implantations were performed with two TIMEs in a fully trained animal. The long term implantations were a success with respect to the fact that the animal recovered quickly with a total absence of paralysis and/or lack of function. Unfortunately the electrode lifetime was rather limited. In the first implantation the median and ulnar TIMEs lasted 2 and 5 months, respectively. In the second implantation they lasted only 2 and 3 weeks. During the long term implantations the TIME’s ability to record neural activity from the median and ulnar nerve was tested, as well as the ability to stimulate the nerve to evoke sensory percepts. This was done in two distinct behavioural tasks. The first was a motor decoding task, in which the animal grasped and lifted a wide variety of objects on a turn table, while simultaneously the neural signals were recorded and the hand kinematics were tracked. After the recording period the animal performed a somatosensory discrimination task with either tactile cues applied to the hand or electrical stimulation to the nerves. The signal-to-noise ratio of the neural recordings was poor and in only a few recordings we were able to detect spiking activity. However it was too sparse for successful decoding of the performed grip type. The long electrode cable under the skin in combination with a dynamic task design introduced too many movement artefacts in the signal. The short lifetime of the electrodes also affected the ability to train the somatosensory discrimination task with electrical stimulation to the nerves. The animal was successfully trained in both the motor decoding task and the somatosensory discrimination task with tactile stimulation. To continue this line of research it would be necessary to move towards a solution with at least an implantable amplifier close to the recording site and preferably also be completely wireless. This would greatly improve the signal-to-noise ratio in the neural recordings and thus the ability to detect and decode neural activity. The TIME in its current form is not stable enough for long term implantation and thus for investigating somatosensory stimulation. Last but not least, while the macaque model is sufficient for basic research and the establishment of stimulation methods, the more detailed exploration of somatosensory restoration (such as different sensory percepts) will necessarily require to move to human subjects (or patients) in order to obtain oral feedback about the elicited percepts.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have