Evaluation of Arm Swing Features and Asymmetry during Gait in Parkinson's Disease Using the Azure Kinect Sensor

Abstract

Arm swinging is a typical feature of human walking: Continuous and rhythmic movement of the upper limbs is important to ensure postural stability and walking efficiency. However, several factors can interfere with arm swings, making walking more risky and unstable: These include aging, neurological diseases, hemiplegia, and other comorbidities that affect motor control and coordination. Objective assessment of arm swings during walking could play a role in preventing adverse consequences, allowing appropriate treatments and rehabilitation protocols to be activated for recovery and improvement. This paper presents a system for gait analysis based on Microsoft Azure Kinect DK sensor and its body-tracking algorithm: It allows noninvasive full-body tracking, thus enabling simultaneous analysis of different aspects of walking, including arm swing characteristics. Sixteen subjects with Parkinson's disease and 13 healthy controls were recruited with the aim of evaluating differences in arm swing features and correlating them with traditional gait parameters. Preliminary results show significant differences between the two groups and a strong correlation between the parameters. The study thus highlights the ability of the proposed system to quantify arm swing features, thus offering a simple tool to provide a more comprehensive gait assessment.

Keywords: Azure Kinect; Parkinson’s disease; arm swing; asymmetry; center of mass sway; gait analysis; movement analysis; spatiotemporal parameters.

Figure 1

(a) The proposed system with the mini-PC, Azure Kinect, and a monitor; (b) GUI for the acquisition of gait trials.

Figure 2

Example of arm swing trajectories (in meters) and gait patterns for PD subjects. (a) Subject #10: arm swing in AP and ML directions for left arm (red line) and right arm (blue line) relative to pelvis joint (black cross); (b) subject #10: gait patterns with detected steps and center-of-mass trajectory within the GAP zone; (c) subject #6: arm swing on AP and ML directions for left arm (red line) and right arm (blue line) relative to pelvis joint (black cross); (d) subject #6: gait patterns with detected steps and center-of-mass trajectory within the GAP zone.

Figure 3

Example of arm swing trajectories (in meters) and gait patterns for HC subjects. (a) Subject #5: arm swing in AP and ML directions for left arm (red line) and right arm (blue line) relative to pelvis joint (black cross); (b) subject #5: gait patterns with detected steps and center-of-mass trajectory within the GAP zone; (c) subject #3: arm swing in AP and ML directions for left arm (red line) and right arm (blue line) relative to pelvis joint (black cross); (d) subject #3: gait patterns with detected steps and center-of-mass trajectory within the GAP zone.

Figure 4

Spearman’s correlation coefficients between parameters for HC and PD groups. Colored boxes indicate the level of significance: yellow (p < 0.05); orange (p < 0.01); red (p < 0.001).