Toward Universal Neural Interfaces for Daily Use: Decoding the Neural Drive to Muscles Generalises Highly Accurate Finger Task Identification Across Humans

Abstract

Peripheral neural signals can be used to estimate movement-specific muscle activation patterns for the purpose of human-machine interfacing (HMI). The available HMI solutions, however, provide limited movement decoding accuracy that often results in inadequate device control, especially in the dynamic tasks context, and require extensive algorithm training that is highly subject-specific. Here, we show that dexterous movements can be identified with high accuracy using a physiology-derived and information-theoretically optimised feature space that targets the spatio-temporal properties of the spiking activity of spinal motor neurons (neural features), decomposed from the interference myoelectric signal. Moreover, we show that the movement decoding accuracy based on these neural features is not influenced by the muscle activation level, reaching overall >98% in the full range of forces investigated and from processing intervals as short as 30ms. Finally, we show that the high accuracy in individual finger movement recognition can be achieved without user-specific models. These results are the first to show a highly accurate discrimination of dexterous movement tasks in a wide range of muscle activation levels from near-real time processing intervals, with minimal subject-specific training, and thus are promising for the translation of HMI to daily use. INDEX TERMS dexterous movement classification, Human-Machine Interfaces, information theory, neural drive, universality of neural control

Non-invasive real-time access to the output of the spinal cord via a wrist wearable interface

“…1). The number of identified motoneurons (6 motoneurons per finger contraction at both force efforts) was consistent with the results by Stachaczyk et al (42) who identified between 5-8 motoneurons per finger contraction when recording signals from the forearm flexor muscles. Interestingly, however, the pulse-to-noise ratio levels observed at the wrist in this study were greater than those usually reported for forearm recordings.…”

“…Therefore, the reported accuracy in finger task classification when decoding motoneurons from the wrist is even superior to that of motoneurons decoded from muscle tissue. Stachaczyk et al (42) also found that the neural output was robust to variations in the force level, unlike myoelectric signals from the forearm when predicting finger flexions (42). Indeed, the increased classification error of tendon electric signals when both force levels were combined was in agreement with previous literature on the effect of dynamic contractions in myoelectric pattern recognition (46,47).…”

“…However, no other study has previously addressed the potential of the wrist for noninvasive neural interfacing, thus the only comparative results are from the forearm. In a finger prediction task, Stachaczyk et al (42) obtained similar classification accuracy for the neural output of the spinal cord at the forearm (98%) to the presented here at the wrist (~97%), although only for individual finger tasks (the four digits, excluding the thumb, while here we tested classification over 10 tasks comprising individual and combined finger gestures). Therefore, the reported accuracy in finger task classification when decoding motoneurons from the wrist is even superior to that of motoneurons decoded from muscle tissue.…”

“…This expression can then be inverted by applying linear-instantaneous blind source separation while maximizing the sparseness of the sources, as long as the tendon potential generated by each motoneuron is unique in relation to the potentials generated by other motoneurons. This condition has been extensively validated for the propagating components of the muscle fiber action potentials (37,42,45,49), but it has never been tested for the end-of-fiber components generated at the tendon endings. Therefore, this assumption needed to be confirmed experimentally (see Results).…”

Review of Advanced Medical Telerobots

“…In this case, a brainmachine interface using steady-state visual evoked potentials is introduced to guide the wheelchair while a vision-based algorithm provides simultaneous localization and mapping (SLAM) to help with navigation among the obstacles. Another relevant application is myoelectric control of prostheses [335,336], which allows users to recover lost functionality by controlling a prosthetic robotic device with their remaining muscle activity. In [337], computer vision (for autonomous object recognition) and mechanomyography (to estimate the intended muscle activation) data are fused to conduct a shared control that predicts user intent for grasp and then realizes it.…”