Hybrid Bionic Nerve Interface for Application in Bionic Limbs

Abstract Intuitive and perceptual neuroprosthetic systems require a high degree of neural control and a variety of sensory feedback, but reliable neural interfaces for long‐term use that maintain their functionality are limited. Here, a novel hybrid bionic interface is presented, fabricated by integrating a biological interface (regenerative peripheral nerve interface (RPNI)) and a peripheral neural interface to enhance the neural interface performance between a nerve and bionic limbs. This interface utilizes a shape memory polymer buckle that can be easily implanted on a severed nerve and make contact with both the nerve and the muscle graft after RPNI formation. It is demonstrated that this interface can simultaneously record different signal information via the RPNI and the nerve, as well as stimulate them separately, inducing different responses. Furthermore, it is shown that this interface can record naturally evoked signals from a walking rabbit and use them to control a robotic leg. The long‐term functionality and biocompatibility of this interface in rabbits are evaluated for up to 29 weeks, confirming its promising potential for enhancing prosthetic control.


Experimental section
Neural prosthetic control through electrically evoked BI-R signals: After hook electrodes were implanted in the sciatic nerve, the efferent RPNI signals induced from the BI-R of the bionic interface were recorded in real time through electrical stimulation.The stimulation parameters were a pulse width of 50 µs and an interpulse delay of 10 µs, and a total of 10 stimulations were performed with an intensity of 150 μA.
At the same time, the recorded signal was smoothed by passing through a bandpass filter with a cutoff frequency of 1 to 3500 Hz and a Savitzky-Golay filter to remove noise and to drive the lower limb prosthetics when the evoked RPNI muscle signal was over the threshold (30σ_noise).The RMS of the noise was calculated for 5 seconds of recording in the absence of any stimulus.In addition, to check whether negative working is possible when the bionic interface is used to obtain input data for neural prosthetic control, the stimulation intensity was lowered to 130 μA so that the RPNI signal below the threshold was induced, and a total of 10 simulations were performed.Differential bipolar recording was performed using the two channels located on the BI-R as the working and reference electrodes and the stainless-steel wire attached to the rabbit ear as the ground electrode.Signals were recorded in real time through an electrophysiology amplifier (RHD 2216; Intan Technology) (Figure S9).

Result & Discussion
Neural prosthetic control through electrically evoked BI-R signals: In both positive and negative tests, it was confirmed that the control of the neural prosthesis was possible with 100% accuracy through the electrically induced BI-R signal (Table S2).It was confirmed that the neural prosthesis worked with all BI-R signals induced above the threshold value, and the neural prosthesis did not operate with BI-R signals induced below the threshold value.These results demonstrate the potential for neural prosthetic control through bionic interfaces.
Further experiments on controlling neural prostheses using both BI-R and BI-N signals as well as naturally evoked somatic signals are needed to further demonstrate the potential of the bionic interface.

Figure S1 .
Figure S1.Buckle interface packaging.(a) Schematic diagram of ACF bonding for wiring the buckle interface.(b) Schematic diagram of wire packaging for head-port connection.For data acquisition during chronic implantation, a 65 cm long wire was connected to the head-port and buckle interface.To ensure durability during chronic implantation, the wire was packaged in a helical structure to withstand tension.(c) Completed buckle interface package and wire with helical core structure.

Figure S2 .
Figure S2.Result of the packaged wire strain test.(a) Tensile strength test and (b) strainimpedance test of the packaged wire.The packaged wire was stretched up to 110% through a tensile compression stand (MultiTest-dv; Mecmesin).Impedance measurement of each channel of the packaged wire for each tensile rate showed stable impedance even at 110% tensile rate.

Figure S3 .
Figure S3.Iridium oxide (IrO2) coating.(a) Schematic diagram of the iridium oxide coating method and comparison of electrodes before and after coating.(b) Impedance changes after IrO2 coating.(c) CSC change after IrO2 coating.

Figure S4 .
Figure S4.Surgical procedure for the formation of a hybrid bionic nerve interface.

Figure S6 .
Figure S6.Results of 4-week efferent signal acquisition and classification of RPNI and nerve signals.Neural signals and RPNI signals were separated based on the number of peaks that could be specified excluding stimulus artifacts (number of peaks in the RPNI signal: 1) number of peaks in the neural signal: 2) and amplitude.(a) 150 μA.(b) 200 μA.(c) 250 μA.

Figure S7 .
Figure S7.BI-R neural signal according to the stimulation intensity recorded at 20 weeks after implantation.(a) Signals recorded at the BI-R evoked by increasing the stimulation intensity from 10 μA to 300 μA.(b) Amplitude change according to the stimulus intensity.(c) Latency changes according to the stimulus intensity.(d) Signals recorded at the BI-R evoked by decreasing the stimulation intensity from 300 μA to 10 μA.(e) Amplitude change according to the stimulus intensity.(f) Latency changes according to the stimulus intensity.

Figure S8 .
Figure S8.Recording configuration for neural signal acquisition.Electrically evoked neural signals were induced by stimulation through cuff electrodes placed on the proximal nerve of the RPNI.(a) Recording configuration for differential bipolar signal acquisition.(b) Schematic diagram of neural signal acquisition.

Figure S9 .
Figure S9.Schematic diagram of neural prosthesis control through electrically evoked BI-R signals.

Table S1 .
SNR of neural signals evoked from chronic implantation.

Table S2 .
Results of neural prosthesis control by electrically induced BI-R signals.