Jenny Ji Hyun Kim

Rehabilitation Engineering | Precision Medicine

Recover-on-Track: A Video and LiDAR-based Tool for Stroke Rehabilitation

[2022 – 2023]

Motion Analysis Laboratory, Spaulding Rehabilitation Hospital

Harvard Medical School

Principal Investigator: Paolo Bonato, PhD

Presented at IEEE BSN 2023, Best Poster Award

ABSTRACT

With the rapid expansion of the elderly population and an ongoing increase in life expectancy, innovative strategies for managing geriatric conditions such as stroke have become indispensable. Here, we introduce ’Recover-on-Track’, a telerehabilitation platform that leverages LiDAR sensors and human pose estimation (HPE) algorithms to provide detailed insights into home-based exercises of stroke patients. We integrated 2D HPE with LiDAR depth data to track motion, achieving results comparable to those of existing wearables-based systems. The preliminary analysis shows promising potential to distinguish between impairment levels. Recover-on-Track lays the groundwork for future exploration in telemedicine research, offering potential avenues for personalizing and enhancing stroke rehabilitation.

CONTRIBUTIONS

I led this project from its inception under the supervision of Dr. Paolo Bonato and Dr. Sunghoon Ivan Lee, collaborating with clinicians to evaluate the efficacy of depth-camera and human pose estimation algorithms in tele-rehabilitation. This involved a comprehensive process of evaluating, selecting, and validating various technologies and algorithms. Additionally, I was responsible for establishing an operational pipeline, drafting an Institutional Review Board (IRB) protocol, and collecting preliminary data to support our findings.


Technologies evaluated: Microsoft Kinect, Intel RealSense D415, OAK-D Pro, Apple iPad Pro

Algorithms evaluated: OpenPose, Google’s MediaPipe, V2V-PoseNet

Validation technology: Vicon Motion Capture, Xsens MVN Awinda (wearable sensors)

MOTIVATION

Working within a rehabilitation hospital, I became aware of the urgency and significance of rehabilitative exercises for stroke survivors. It was evident that clinicians tirelessly balanced the quantity of care, measured by the number of patients seen, with the quality of care, reflected in the time spent with each patient. This experience highlighted the potential of technology to provide clinicians and stroke survivors with deeper insights into individual rehabilitation processes. Such technological integration could foster motivation, improve therapy adherence, and increase training volume for patients; additionally, greater insights into the strengths and weaknesses and therapy progression of each patient, this tool would allow clinicians to better tailor the therapy program to each individual.

Motivated by this, I aimed to utilize readily available technologies, like the Apple iPad, along with efficient frameworks like Google’s Mediapipe, to establish the groundwork for a tele-rehabilitation platform specifically designed for stroke survivors.

Made in Palo Alto with lots of Sunshine and Coffee