Kinect-based assessment of proximal arm non-use after a stroke

Area of Science:

• Rehabilitation Medicine
• Biomedical Engineering
• Neuroscience

Background:

• Stroke survivors often exhibit proximal arm non-use (PANU) during seated reaching, compensating with trunk movements.
• Previous quantification of PANU relied on motion capture systems like Zebris CMS20s.

Purpose of the Study:

• To validate a low-cost Microsoft Kinect-based system for diagnosing PANU.
• To compare Kinect's accuracy and reliability against the established CMS20s system.

Main Methods:

• Nineteen hemiparetic stroke individuals participated.
• Measurements of PANU score, reach length, trunk length, and proximal arm use (PAU) were taken simultaneously using Kinect (v2) and CMS20s.
• Two testing sessions were conducted with a two-hour interval.

Main Results:

• Kinect demonstrated strong correlations with CMS20s for PANU score (ICC=0.96), reach length (ICC=0.81), trunk length (ICC=0.97), and PAU (ICC=0.97).
• Both systems showed good test-retest reliability for PANU scores (Kinect: ICC=0.76, CMS20s: ICC=0.72).
• Slight practice effects were observed, with reduced PANU scores in the re-test session for both systems.

Conclusions:

• The Microsoft Kinect system accurately and reliably assesses PANU, reach length, trunk length, and PAU in post-stroke individuals.
• Kinect offers a cost-effective and accessible solution for clinical PANU assessment and monitoring paretic arm recovery progress.