Direct Prosthesis Force Control with Tactile Feedback May Connect with the Internal Model.

INTRODUCTION

Dynamic modulation of grip occurs mainly within the major structures of the brain stem, in parallel with cortical control. This basic, but fundamental level of the brain, is robust to ill-formed feedback and to be useful, it may not require all the perceptual information of feedback we are consciously aware. This makes it viable candidate for using peripheral nerve stimulation (PNS), a form of tactile feedback that conveys intensity and location information of touch well but does not currently reproduce other qualities of natural touch. Previous studies indicate that PNS can integrate with the basic levels of the motor system at a pre-perceptual level and can be processed optimally in multisensory integration, but there is little evidence if PNS is used effectively for motor corrections.

METHODS

We performed a study with an individual with a mid-radial upper limb difference who has cuff electrodes on his peripheral nerves to give him the sense of touch to perform an object movement over a barrier task. During this task we measured how the participant moved the object with a prosthetic hand in space, how they varied their grip force on the object, and how their muscle signals varied as force changed. We tested this with four different conditions: with and without stimulated tactile sensation combined with the user having control over force on an object or velocity of hand movement.

RESULTS

Given direct control of force, the participant's output force significantly correlated with the puck's displacement up to the apex of the movement, but did not correlate afterwards. This indicated a trend of increasing force when lifting the puck, but no decrease when lowering it. In comparison, when the participant moved the puck with the intact hand, they had a small but significant increase in force when lifting the puck in half the cases, but always had a significantly decrease in force when lowering the puck. When the participant used a force controller with stimulation, the puck slipped or dropped significantly more times (p < 0.05) compared to the velocity controller with stimulated feedback. This result implied that when the participant intended to loosen their grip, the prosthesis opened instead, which would explain the lack of force reduction in the initial results. The analysis of intent decoded from EMG during use of the force controller shows that the participant intended to lower their grip force with or without stimulation when using a high shatter threshold, but when using a lower threshold, the stimulation gave the participant a better sense of where the shatter threshold was. With a low shatter force, the participant tended to modulate their muscle contractions to a constant level if they were given stimulation (no significant correlation with movement) or they generally increased their intended force towards the shatter force threshold without stimulated feedback. With a moderate shatter force, the participant kept a relatively constant contractile force with or without stimulation. In contrast the EMG analysis with the velocity controller has a mixed trend of increasing and decreasing muscle indicating no global desire to change their grip force in one direction or the other. Finally, analysis of the puck movement showed that the participants moved the puck higher above the barrier with the force controller compared to movements with the velocity controller (p < 0.001), but the addition of stimulation with either controller lowered the participant's movements significantly closer to the barrier (p < 0.001). Stimulation may cause an instantaneous increase in confidence with a controller or create better positional awareness with either controller.

DISCUSSION

While the participant of this study did not show any significant output grip force changes during the object movement tasks, their decoded intent combined with the higher number of loosening events when using the force controller and with stimulation indicates they may have been trying to reduce their grip force during the task. This behavior matches with the force output of the participant's intact hand. In order convert the participant's intent into the correct output force, there needs to be changes to the overall design of modern prosthetic devices to allow for smaller grip force changes and changes to force within a static grip. Furthermore, improvements to the stimulation that amplify small changes in force and estimate the any slip forces on the fingertips will provide more useful signals to the participant.