Intelligent Evaluation Method of Human Cervical Vertebra Rehabilitation Based on Computer Vision

Area of Science:

• Biomedical Engineering
• Computer Vision
• Rehabilitation Science

Background:

• Rising incidence of cervical spondylosis, particularly in adolescents, due to modern work and lifestyle changes.
• Lack of mature, unmanned systems for evaluating and monitoring cervical spine rehabilitation exercises.
• Risk of patient injury due to insufficient physician guidance during exercises.

Purpose of the Study:

• To propose a novel computer vision-based method for assessing cervical spine exercises.
• To develop an unmanned system for guiding and evaluating patient rehabilitation.
• To enhance the safety and effectiveness of cervical spine disease rehabilitation.

Main Methods:

• Utilizing the Mediapipe framework to create a face mesh and extract features for calculating 3-DOF head pose angles.
• Calculating sequential angular velocity in 3-DOF from acquired computer vision data.
• Developing a privacy encryption algorithm (YOLOv5 and mosaic noise mixing) for head posture information.

Main Results:

• The proposed algorithm demonstrates good repeatability in assessing cervical spine exercises.
• The system effectively reflects the health status of the patient's cervical spine.
• Successful implementation of a privacy-preserving system for remote rehabilitation monitoring.

Conclusions:

• The developed computer vision algorithm provides a viable alternative to physician guidance for cervical spine rehabilitation.
• The system offers a reliable and repeatable method for evaluating exercise performance and patient health.
• The integration of privacy protection ensures secure and confidential patient data handling.