NASA’s new dark energy space telescope can also detect killer asteroids
technologyreview.com·1h ago
AI Research
arXiv CS.AI·Jun 9·5 min read
Quantifying Rodda and Graham Gait Classification from 3D Markerless Kinematics derived from a Single-view Video in a Heterogeneous Pediatric Clinical Cohort
TL;DR
Cerebral Palsy (CP) affects children's movement, making accurate gait assessment crucial for maintaining walking ability. A method was developed to classify gait deviations using 3D markerless kinematics from a single-view video, applying the Rodda and Graham classification system.
✦ Why It Matters
Engineers can develop video-based tools for gait analysis, improving accessibility and accuracy in clinical assessments.
Key Takeaways
- Utilize single-view videos for gait analysis in pediatric patients.
- Achieve high accuracy in knee z-score estimation with R^2 of 0.80.
- Identify excess knee flexion effectively with 83% accuracy.
- Support longitudinal tracking of gait changes over time.
How It Works
The markerless gait analysis pipeline uses computer vision techniques to analyze single-view videos, extracting 3D kinematic data without the need for physical markers. It calculates knee and ankle z-scores by comparing the observed gait patterns to established norms, enabling accurate classification of gait deviations.