An Engineered Tactile Modulation Platform to Recreate Tactile Feedback in Upper-Limb Prostheses

The lack of tactile feedback in grasping objects is one barrier preventing the widespread adoption of upper-limb motorized prostheses. In its absence, amputee users visually attend to graspers, but this creates a high cognitive demand. Efforts to restore tactile sensation require nerve interfaces that use artificial force sensors and electrical stimulation to evoke ionic biological responses. The work herein contributes to the design and testing of a tactile modulation platform that employs models of biological mechanotransduction to better translate force applied at a sensor into biphasic current pulses to be delivered to peripheral afferents, which is done at present in an ad hoc fashion. In specific, the platform consists of a leaky-integrate-and-fire software algorithm of the neuron that generates the timing of the trains of biphasic current pulses, and custom hardware circuitry to control the amplitude and duration of individual pulses. Because of known, naturally occurring phenomenon in the nervous system as well as man-made sensors and electrodes, the platform’s parameters need to be calibrated to enable long-term use. To achieve this, procedures and a user interface were designed for use with a mobile phone to afford an intuitive means of adjustment for amputee users with no means to understand the mathematics. Experimental studies were conducted to compare the platform’s predicted firing rate to prior electrophysiological recordings in the sural nerve of the rat by applying three levels of force (1.5, 4, 7.5 N), in the nominal case. Then, parameters related to i) nerve discrimination sensitivity were varied for comparison to recorded data for three rats with different stimulus-response transformations, ii) sensor gain were adjusted to account for the variability among three force sensors, and iii) nerve absolute threshold were varied to achieve different biphasic pulse amplitudes. The results indicate that firing rates generated compare favorably to those observed, for both the 0.5 sec duration dynamic ramp-up (11.2, 28.5, to 42.9 spikes/sec) and the static hold between 2 and 5 sec of the stimulus (7.5, 14.5, to 30.1 spikes/sec). In addition, the platform could account for the both more and less steep sensitivity functions of three additional rats, producing static phase firing rates spanning 6.7 – 129.8 spikes/sec. Three force sensors were standardized to a load cell (standard of comparison) with linear goodness of fit values above 0.9 and estimated force compared to actual force with all root mean square errors below 1. Finally, the platform was able to produce amplitudes of biphasic pulses over a range of ±0.2– 15 V.