Skip to main content

ORIGINAL RESEARCH article

Front. Bioeng. Biotechnol.
Sec. Biomechanics
Volume 12 - 2024 | doi: 10.3389/fbioe.2024.1520831
This article is part of the Research Topic Biomechanics, Sensing and Bio-inspired Control in Rehabilitation and Assistive Robotics, Volume II View all articles

A Novel Multi-Level 3D Pose Estimation Framework for Gait Detection of Parkinson's Disease Using Monocular Video

Provisionally accepted
He Rong He Rong 1Zijing You Zijing You 2,3Yongqiang Zhou Yongqiang Zhou 2*Guilan Chen Guilan Chen 2*Yanan Diao Yanan Diao 2Xiantai Jiang Xiantai Jiang 2*Yunkun Ning Yunkun Ning 2*Guoru Zhao Guoru Zhao 2*Ying Liu Ying Liu 1*
  • 1 Department of Rehabilitation Medicine, University of Hong Kong-Shenzhen Hospital, Shenzhen, China
  • 2 Key Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (CAS), Shenzhen, China
  • 3 University of Chinese Academy of Sciences, Beijing, Beijing, China

The final, formatted version of the article will be published soon.

    Parkinson's disease (PD) patients show muscle stiffness, bradykinesia, and balance disorders, which seriously reduce the patient's quality of life. Motion pose estimation and gait analysis can help patients with early diagnosis and timely intervention, but clinical practice lacks objective and accurate gait analysis tools. This work proposes a multi-level 3D pose estimation framework for PD patients based on monocular video combined with Transformer and graph convolutional network (GCN) frameworks. The repeatability and effectiveness of gait temporal and spatial parameters of 59 healthy elderly and PD patients were extracted and verified, and an early prediction model for PD patients was established. The results showed that most of the repeatability results of the estimated parameters were ICC (2, k) > 0.70, and most of the parameters were highly correlated with VICON and ATMI results (r > 0.80); in addition, the classification results based on the proposed parameter features and Random Forest (RF) classifier had an accuracy of 93.3%. The PD pose estimation method proposed in this study can provide reliable and effective 3D human pose parameters and has good feasibility in early prediction. The markerless 3D human pose estimation method proposed in this study has the advantages of low cost, portability, and simple operation, and is expected to become a new tool for monitoring and screening PD patients that is widely used in clinical practice.

    Keywords: Parkinson's disease (PD), 3D pose estimation, monocular video, Graph convolutional network (GCN), Gait detection

    Received: 31 Oct 2024; Accepted: 29 Nov 2024.

    Copyright: © 2024 Rong, You, Zhou, Chen, Diao, Jiang, Ning, Zhao and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

    * Correspondence:
    Yongqiang Zhou, Key Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (CAS), Shenzhen, China
    Guilan Chen, Key Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (CAS), Shenzhen, China
    Xiantai Jiang, Key Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (CAS), Shenzhen, China
    Yunkun Ning, Key Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (CAS), Shenzhen, China
    Guoru Zhao, Key Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (CAS), Shenzhen, China
    Ying Liu, Department of Rehabilitation Medicine, University of Hong Kong-Shenzhen Hospital, Shenzhen, China

    Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.