Estimation of the Knee Joint with Single-Camera Smartphone

Area of Science:

• Biomechanics
• Computer Vision
• Rehabilitation Technology

Background:

• Gait analysis quantifies walking patterns, crucial for rehabilitation.
• Markerless motion capture offers a low-cost, portable alternative.
• Computer vision techniques enable real-time gait assessment.

Purpose of the Study:

• To evaluate a smartphone-based gait analysis system.
• To assess the accuracy of knee and ankle kinematics.
• To validate the system against wearable sensors.

Main Methods:

• Utilized a single smartphone camera and MediaPipe pose estimation.
• Assessed knee and ankle kinematics in 27 healthy volunteers.
• Validated results against the OPAL wearable sensor system.

Main Results:

• High accuracy for knee flexion, extension, and range of motion (low error, high correlation).
• Excellent agreement for knee kinematics between the smartphone and wearable systems.
• Poor concordance for ankle kinematics, indicating significant differences and errors.

Conclusions:

• Smartphone-based gait analysis is effective and affordable for knee kinematics.
• The system provides real-time, portable assessment capabilities.
• Ankle motion analysis using this method requires further refinement.