Quantifying Prosthetic and Intact Limb Use in Upper Limb Amputees via Egocentric Video: An Unsupervised, At-Home Study

Analysis of the manipulation strategies employed by upper-limb prosthetic device users can provide valuable insights into the shortcomings of current prosthetic technology or therapeutic interventions. Typically, this problem has been approached with survey or lab-based studies, whose prehensile-gra...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on medical robotics and bionics 2021-05, Vol.3 (2), p.463-484
Main Authors: Spiers, Adam J., Cochran, Jillian, Resnik, Linda, Dollar, Aaron M.
Format: Article
Language:English
Subjects:
Amputation
Atmospheric measurements
Cameras
End effector
human manipulation
human motion analysis
Myoelectricity
Particle measurements
Prostheses
prosthetic hands
Prosthetics
Tagging
Task analysis
Taxonomy
upper limb prosthetics
Video data
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Analysis of the manipulation strategies employed by upper-limb prosthetic device users can provide valuable insights into the shortcomings of current prosthetic technology or therapeutic interventions. Typically, this problem has been approached with survey or lab-based studies, whose prehensile-grasp-focused results do not necessarily give accurate representations of daily activity. In this work, we capture prosthesis-user behavior in the unstructured and familiar environments of the participants own homes. Compact head-mounted video cameras recorded ego-centric views of the hands during self-selected household chores. Over 60 hours of video was recorded from 8 persons with unilateral amputation or limb difference (6 transradial, 1 transhumeral, 1 shoulder). Of this, almost 16 hours of video data was analyzed by human experts using the 22-category 'TULIP' custom manipulation taxonomy, producing the type and duration of over 27,000 prehensile and non-prehensile manipulation tags on both upper limbs, permitting a level of objective analysis not previously possible with this population. Our analysis included unique observations on non-prehensile manipulations occurrence, determining that 79% of transradial body-powered device manipulations were non-prehensile, compared to 60% for transradial myoelectric devices. Conversely, only 16-19% of intact limb activity was non-prehensile. Additionally, multi-grasp terminal devices did not lead to increased activity compared to 1DOF devices.
ISSN:2576-3202
2576-3202
DOI:10.1109/TMRB.2021.3072253