UW Researcher Explores Home-Based Telehealth Webcam Setup to Measure Joint Range of Motion

The COVID-19 pandemic brought into focus the need for more and better telehealth medicine, especially in a state such as Wyoming, where rural towns and cities are separated by 50-100 miles.

A researcher in the University of Wyoming College of Health Sciences led a recent study that looked at a potentially more accurate and less expensive way -- using a webcam setup at home -- to measure, in 3D, the range of motion of multiple joints during physical therapy health care conducted remotely.

“The crux of this study is that a webcam -- costing less than $100, easy to set up at home and enhanced by machine learning -- can replace the 3D motion capture system -- costing more than $100,000 and requiring professional installation at a research lab -- to perform a decent job in evaluating joint range-of-motion evaluation for those physical therapists who are practicing in telehealth mode,” says Qin Zhu, a UW professor of kinesiology and health.

Zhu was senior and co-author of a paper titled “A webcam-based machine learning approach for three-dimensional range of motion evaluation” that was published Oct. 23 in PLOS ONE, an inclusive journal community working together to advance science for the benefit of all society, now and in the future. 

In the context of physical therapy and rehabilitation, joint range of motion is a widely used outcome measure. Previous body joint range-of-motion studies commonly relied on the use of a hand-held goniometer, which requires certified physicians or physical therapists with significant training and practice using the device. This imposes challenges to medically underserved communities -- rural and remote -- where access to health care and trained professionals is limited, according to the paper.

This new study presents and evaluates an alternative machine learning-based, range-of-motion evaluation method that can be remotely accessed via a webcam. Range of motion is commonly used to evaluate and quantify the improvement in a patient’s mobility at a certain joint because of an intervention.

“In a rural state like Wyoming, telehealth is sought by both physical therapists and patients to reduce travel and costs associated with in-person visits to clinics while improving therapist-patient contact time and patient adherence to the therapy,” Zhu says. “Now, the physical therapist can remotely assess the patient’s motor functions by asking the patient to set up a webcam at home recording their movements. Compared to the traditional measure of range of motion using a goniometer, the webcam image-based solution is more objective and accurate, and hands-free.”

Twenty-five healthy adults -- 13 females and 12 males from the UW Division of Kinesiology and Health -- participated in the study, which took place between March-November 2022. Participants wore black motion-capture suits fitted with 39 reflective markers.

A program, called BlazePose, was used to generate 3D coordinates of 33 different joints. To derive the range-of-motion angles for different joints, joint trajectories of various range-of-motion evaluation movements were simultaneously recorded using a webcam-based, real-time pose estimation algorithm and a marker-based infrared optical motion capture system.

Movements of the spine (extension and flexion, lateral flexion and trunk rotation); neck (extension and flexion, lateral bending and rotation); upper extremities (shoulder adduction and abduction, shoulder extension and flexion, and elbow extension and flexion); and lower extremities (hip extension and flexion with knee extended, hip flexion with knee flexed, and hip adduction and abduction) were recorded. Movements of distal joints, such as the wrist and ankle, were excluded from the study because BlazePose could not reliably generate marker positions of the hands and feet, respectively, according to the study.

This technology can be easily implemented for telerehabilitation, reducing the costs and increasing medical accessibility for patients, Zhu says.

“Not only can range of motion be measured, but other aspects of motion analysis -- reaction time, movement time, speed/accuracy/trajectory of movement, interjoint coordination and hand dexterity -- can be performed as well,” he says.

The current study stresses subjects were healthy adults who could perform various range-of-motion movements in a standing posture. However, in practical applications, this tool will be used by patients who may have physical frailty or neurodivergent conditions, making it challenging for them to achieve the standing position used in the study.

Zhu admits there are limits to the study. In addition to the system’s unreliability in measuring movement of distal joints, he says the accuracy of tracking and range-of-motion estimation using the webcam-based approach was not 100 percent equivalent to the “golden standard 3D motion capture system” in which accuracy is validated to be high.

Xiaoye Michael Wang, a research associate at the University of Toronto, was lead and corresponding author of the paper. Derek Smith, an associate professor and director of UW’s Division of Kinesiology and Health, was the second author of the paper.

About UW’s College of Health Sciences

UW’s College of Health Sciences trains health and wellness professionals and researchers in a wide variety of disciplines, including medicine, nursing, pharmacy, speech-language pathology, social work, kinesiology, public health, health administration and disability studies. The college also oversees residency and fellowship programs in Casper and Cheyenne, as well as operating a speech/hearing clinic in Laramie and primary care clinics in Laramie, Casper and Cheyenne.

With more than 1,600 undergraduate, graduate and professional students, the college is dedicated to training the health and wellness workforce of Wyoming and conducting high-quality research and community engagement, with a particular focus on rural and frontier populations.