Spatiotemporal Gait Measurement With a Side-View Depth Sensor Using Human Joint Proposals.

Area of Science:

• Biomechanics
• Human Motion Analysis
• Computer Vision

Background:

• Accurate measurement of gait parameters is crucial for clinical assessments.
• Existing side-view systems often rely on indirect measures, limiting precision.
• Depth sensors offer a non-invasive approach to capturing human movement.

Purpose of the Study:

• To develop and validate a method for calculating standard spatiotemporal gait parameters using a side-view depth sensor.
• To directly derive gait metrics from individual joint positions.
• To compare the accuracy of the proposed method against a gold standard pressure-sensitive walkway.

Main Methods:

• Utilized a side-view Kinect depth sensor to capture clinical walking trials.
• Employed a stochastic predictor to generate joint proposals from depth images, represented as a weighted graph.
• Determined accurate foot positions by comparing shortest paths through the graph.
• Detected stance phases from foot motion over time to calculate gait parameters.
• Validated results using Bland-Altman analysis and intraclass correlation coefficients against the Zeno Walkway.

Main Results:

• The proposed system achieved high agreement with the Zeno Walkway for spatial gait parameters.
• Intraclass correlation coefficients (ICC) demonstrated strong agreement, with ICC(2, 1) = 1.00 for stride length and 0.98 for step length at normal pace.
• The method directly calculates gait parameters from foot positions, unlike previous indirect approaches.
• The side-view system enables bidirectional walking tracking within a single camera's field of view.

Conclusions:

• The developed method provides accurate and direct calculation of spatiotemporal gait parameters from individual joint data using a side-view depth sensor.
• This approach offers a promising, non-invasive tool for gait analysis in clinical and research settings.
• The system's ability to track bidirectional walking enhances its applicability in various environments.