Testing the feasibility of quantizing the progress of stroke patients’ rehabilitation with a computer vision method

Shuai Zhao, Yao Tong, Gang Chen, Yiran ZhangICCV 2022 · DOI

Abstract

Limb rehabilitation training facilitates hemiparalysis stroke patients’ recovery and improves their quality of life. Knowing the rehabilitation status is essential for both physicians and patients. Instead of relying on wearable sensors or deep cameras, the computer vision method can recognize a patient’s training action, movement trajectory, and activity status, which provides a more accurate and usability path to measure the rehabilitation. However, except for static measures of real-time behavior, quantizing the dynamic change of varying training periods to evaluate the rehabilitation progress is nontrivial in the clinic. In this paper, we proposed a computable method to compare the motion change of the upper limb. We first used OpenPose to preprocess the video data to recognize the joint points of the upper limb, and described the position of each joint point by Cartesian coordinates. Second, we used the dynamic time warping algorithm to compute the similarity of the limb’s lift angle and time in varying training periods to obtain the rehabilitation progress. The results show that our method can achieve data measurement and analysis of the quality of rehabilitation actions by a simple monocular camera, which has a good application prospect.
BibTeX
@inproceedings{zhao2022testing,
  author = {Shuai Zhao and Yao Tong and Gang Chen and Yiran Zhang},
  title = {{Testing the feasibility of quantizing the progress of stroke patients’ rehabilitation with a computer vision method}},
  booktitle = {IEEE/CVF International Conference on Computer Vision},
  year = {2022},
  doi = {10.1109/cvidliccea56201.2022.9824678},
}

Loading related content...

Figure