Egocentric video: a new tool for capturing hand use of individuals with spinal cord injury at home

Journal of NeuroEngineering and Rehabilitation, 2019 · DOI: https://doi.org/10.1186/s12984-019-0557-1 · Published: June 25, 2019

Spinal Cord Injury Assistive Technology Biomedical

Simple Explanation

This study introduces a wearable camera system to track hand use in people with spinal cord injuries at home. The system uses computer vision to identify the hands and their interactions with objects. The system's algorithm detects the hand, segments it from the background, identifies left or right hand, and detects functional interactions with objects. The algorithm's accuracy was tested against manual video analysis, showing promising results in capturing hand-object interactions, paving the way for improved home-based assessments.

Study Duration

Not specified

Participants

9 participants with cSCI for testing, 8 participants with cSCI for training

Evidence Level

Not specified

Key Findings

1
The algorithm achieved F1-scores of 0.74 ± 0.15 for the left hand and 0.73 ± 0.15 for the right hand when compared to manual video labelling.
2
Moderate and significant correlations were found between the automated system's output and manual labelling for total interaction time, average interaction duration, and interactions per hour.
3
Feature analysis showed that optical flow, hand shape (HOG), and colour histograms all contribute to hand-object interaction classification, with combined features providing the highest performance.

Research Summary

This study presents a wearable egocentric camera system for capturing quantitative measures of hand use at home for individuals with cervical spinal cord injury (cSCI). The system uses computer vision algorithms to detect and segment the hand, distinguish between left and right hands, and detect functional interactions with objects during activities of daily living (ADLs). The results demonstrate the potential of this wearable camera system for providing objective, quantitative measures of hand function in the home environment, addressing a gap in current assessment methods.

Practical Implications

Home-Based Assessment

Enables objective, quantitative assessment of hand function in real-world environments, addressing the limitations of clinic-based assessments.

Outcome Measure Development

Provides a foundation for developing novel outcome measures that capture real-world hand use, useful for evaluating rehabilitation interventions.

Personalized Rehabilitation

Facilitates personalized rehabilitation programs by providing detailed insights into an individual's hand use patterns in their daily lives.

Study Limitations

1
The performance of the hand-object interaction system is dependent on the accuracy of the hand detection and segmentation steps.
2
The algorithm remains computationally expensive for a mobile system.
3
The algorithms need to be evaluated in a wider range of environments, with challenges that may include imperfect lighting, differences in camera orientation, and more diverse tasks.

Your Feedback

Was this summary helpful?

Back to Spinal Cord Injury