Quantitative Comparison of Hand Kinematics Measured with a Markerless Commercial Head-Mounted Display and a Marker-Based Motion Capture System in Stroke Survivors

Abstract

Upper-limb paresis is common after stroke. An important tool to assess motor recovery is to use marker-based motion capture systems to measure the kinematic characteristics of patients' movements in ecological scenarios. These systems are, however, very expensive and not readily available for many rehabilitation units. Here, we explored whether the markerless hand motion capabilities of the cost-effective Oculus Quest head-mounted display could be used to provide clinically meaningful measures. A total of 14 stroke patients executed ecologically relevant upper-limb tasks in an immersive virtual environment. During task execution, we recorded their hand movements simultaneously by means of the Oculus Quest's and a marker-based motion capture system. Our results showed that the markerless estimates of the hand position and peak velocity provided by the Oculus Quest were in very close agreement with those provided by a marker-based commercial system with their regression line having a slope close to 1 (maximum distance: mean slope = 0.94 ± 0.1; peak velocity: mean slope = 1.06 ± 0.12). Furthermore, the Oculus Quest had virtually the same sensitivity as that of a commercial system in distinguishing healthy from pathological kinematic measures. The Oculus Quest was as accurate as a commercial marker-based system in measuring clinically meaningful upper-limb kinematic parameters in stroke patients.

Keywords: Oculus Quest; Optitrack; head-mounted display; markerless motion capture system; motion analysis; motor recovery; stroke.

Figure 1

Experimental protocol and examples of hand trajectories and velocity profiles. (A) Perspective (top panel) and top (bottom panel) view of a graphical depiction of the experimental conditions. The patients wore a head-mounted display and were immersed in a virtual environment in which they played a game with motor rehabilitation purposes (i.e., hit virtual glasses and make them fall). During the task, their hand movements were recorded by the Oculus on-board sensors and software and by means of a commercial motion capture system (Optitrack, NaturalPoint, Inc., Corvallis, OR, USA), respectively. The gray circles represent the placements on the patient’s body of the Optitrack markers. The inset in the top panel shows the patient’s view of the virtual environment inside the HMD. (B) Perspective (top panel) and top (bottom panel) view of hand trajectories recorded by the Optitrack and Oculus systems during one experimental session. Left and right hand trajectories are plotted in blue and light brown, respectively. (C) Examples of speed profiles computed from Oculus (solid line) and Optitrack (dashed line) data for the left (top panel) and right (bottom panel) hand, respectively. Further examples of hand speed profiles are shown in Figure S1.

Figure 2

Estimation of distances and velocities by the Optitrack and Oculus systems, respectively. (A) Maximum reaching distances (top) and peak velocities (bottom) measured in an experimental session for one of our patients. In both panels, each dot represents a single trial. (B) Slopes of the linear regression between measures of maximum reaching distance and peal velocity computed from Optitrack and Oculus data. Each dot represents a patient. Mean slope for estimates of maximum reaching distance = 0.94 ± 0.1. Mean slope for estimates of peak velocities = 1.06 ± 0.12. (C) Estimates of maximum reaching distances (left panel) and peak velocity (right panel) computed for all patients from Optitrack and Oculus data. Each dot represents a single trial. In both panels, the red line represents a linear fit of the data. The slopes of this fit is shown in the caption. (D) Bland–Altman plots of the data in panel (C).

Figure 3

Comparison of kinematic assessments using data from the Optitrack and Oculus systems. The two boxplots show the distributions of peak velocities of healthy and impaired hand obtained with the Optitrack (A) and Oculus (B) systems in one patient. In both panels, each dot represents a trial and the asterisk indicates a statistically significant difference (p < 0.05).