Validation of a Kinect V2 based rehabilitation game

Area of Science:

• Rehabilitation Technology
• Biomechanics
• Neurorehabilitation

Background:

• Interactive technologies show promise in stroke rehabilitation but lack validation as assessment tools.
• The Microsoft Kinect V2 is used in the Mystic Isle game for stroke patients' motor function and activity assessment.
• Accuracy of Kinect V2 for upper-body movements requires further validation compared to gold-standard systems.

Purpose of the Study:

• To evaluate the spatial accuracy and measurement validity of the Kinect-based game Mystic Isle.
• To compare Mystic Isle's assessment capabilities against the Vicon optical motion capture system.
• To establish the reliability of Mystic Isle for assessing upper-body movements in stroke patients.

Main Methods:

• Thirty participants performed six trials in sitting and standing positions.
• Simultaneous data collection from Kinect V2 and Vicon systems, followed by filtering and synchronization.
• Spatial accuracy assessed using Pearson's r, SNR, and 3D distance difference; measurement validity evaluated using error metrics and ICC.

Main Results:

• Arm joint signals demonstrated high spatial accuracy with correlation coefficients >0.9 and SNR >5.
• Hip joint data showed lower stability and greater SNR variation.
• Mean 3D joint distance differences were below 10 cm; hand reach and velocity errors were within acceptable ranges (<5% and <10%).

Conclusions:

• Mystic Isle, utilizing Kinect V2, provides accurate movement measurements and assessments for stroke rehabilitation.
• The game demonstrates sufficient spatial accuracy and measurement validity when compared to the Vicon system.
• Findings support the use of Mystic Isle as a reliable tool for assessing upper-body motor function in stroke recovery.