AUTHOR=Perry Briana N. , Armiger Robert S. , Yu Kristin E. , Alattar Ali A. , Moran Courtney W. , Wolde Mikias , McFarland Kayla , Pasquina Paul F. , Tsao Jack W. 

TITLE=Virtual Integration Environment as an Advanced Prosthetic Limb Training Platform

JOURNAL=Frontiers in Neurology

VOLUME=Volume 9 - 2018

YEAR=2018

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2018.00785

DOI=10.3389/fneur.2018.00785

ISSN=1664-2295

ABSTRACT=Background: Despite advances in prosthetic development and neurorehabilitation, individuals with upper extremity (UE) loss continue to face a variety of functional and psychosocial challenges following amputation. Loss of autonomy, inability to perform activities of daily living, and a reduced quality of life are just some of the difficulties that accompany UE amputation. Many individuals with UE loss opt to use a prosthesis, which invites the additional challenges of proper prosthetic selection, fitting, and training. In recent years, advanced myoelectric prostheses have emerged on the marketplace and offer seemingly intuitive manipulation of multiple simultaneous degrees of prosthetic motion, as well as the potential for sensory feedback integration. Successful manipulation of such advanced prostheses, however, requires complex control paradigms and training methods. We explored whether the Virtual Integration Environment (VIE), a virtual reality simulator, could be used to teach dexterous prosthetic control paradigms, such as pattern recognition feedback, to individuals with UE loss.

Methods: Active duty military personnel with UE loss were recruited at Walter Reed National Military Medical Center (WRNMMC) in Bethesda, MD. Thirteen individuals representing 14 study limbs completed 20, 30-minute virtual reality therapy sessions over the course of one to two months. These sessions required the participant to passively follow the upper limb motions of a virtual avatar using their residual and phantom limb. Signal capture from the residual limb was accomplished during passive training using eight pairs of surface electromyography (sEMG) electrodes placed circumferentially around the residual limb. Eight participants, representing nine study limbs, additionally completed 20, 30-minute active motor training and assessment sessions using the MiniVIE platform, which enabled them to drive the virtual avatar through three motion sets – Basic, Advanced, and Digit control. These assessments generated training and motion accuracy scores, which were analyzed over time. Seven of the eight participants completed a weekly control evaluation with their intact limb. The eighth participant had bilateral upper extremity amputations and opted to complete the study twice – once with each upper limb. 
Results: The mean classification accuracy score for the 14 limbs studied across all 277 passive motor control sessions was 43.8 ± 10.7%. There were