Yun-Chien Yeh
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Yun-Chien Yeh has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Cheng-Yu Tsai
Research Center of Sleep Medicine, College of Medicine, Taipei Medical University, Taipei 11031, Taiwan
1 shared publication - Ching-Shiou Tang
School of Biomedical Engineering, Taipei Medical University, Taipei 110, Taiwan
1 shared publication - Hung Quang Ho
Taipei Medical University
1 shared publication - Jiunn-Horng Kang
Department of Physical Medicine and Rehabilitation, Taipei Medical University Hospital, Taipei 110301, Taiwan
1 shared publication
Latest Publications
- Open Access
A Feasible Method for Evaluating Post-Stroke Knee Spasticity: Pose-Estimation-Assisted Pendulum Test
Purpose: Post-stroke spasticity (PSS) substantially affects functional recovery and quality of life in stroke survivors. However, the current clinical assessment methods exhibit certain subjectivity and equipment limitations. Human pose estimation presents a promising alternative for objective and user-friendly spasticity assessment. Materials and Methods: A total of 20 stroke survivors with PSS underwent pendulum tests with smartphones from multiple angles to quantitatively assess knee muscle spasticity. Pose estimation was conducted using the AlphaPose and STCFormer algorithms, with simultaneous measurements using an electronic goniometer as a reference. Three pendulum parameters were evaluated: normalized relaxation index (P1), first maximum of oscillation (P2), and relaxation index at half swing (P3). Bland–Altman analyses were used to analyze the consistency between pose estimation and electronic goniometer measurements. Intraclass correlation coefficient (ICC) and Spearman’s correlation analyses were conducted to evaluate agreement and reliability between electronic goniometer measurements and clinical evaluation. Results: P1 demonstrated the highest consistency between pose estimation and electronic goniometer measurements, with the highest ICC values (0.931 for AlphaPose and 0.911–0.94 for STCFormer). P1 and P3 differentiated between affected and unaffected limbs (p < 0.01) and demonstrated significant negative correlations with Modified Ashworth Scale scores, particularly for knee extensors (P1: ρ = −0.747 for AlphaPose and −0.781 for STCFormer; p < 0.01). P2 demonstrated low consistency and differential performance across all analyses. Conclusions: Video-based human pose estimation, particularly using P1, offers a reliable and objective method for evaluating PSS, demonstrating strong agreement with electronic goniometer measurements. This approach is clinically feasible for evaluating spasticity.
Open AccessA Feasible Method for Evaluating Post-Stroke Knee Spasticity: Pose-Estimation-Assisted Pendulum Test
Purpose: Post-stroke spasticity (PSS) substantially affects functional recovery and quality of life in stroke survivors. However, the current clinical assessment methods exhibit certain subjectivity and equipment limitations. Human pose estimation presents a promising alternative for objective and user-friendly spasticity assessment. Materials and Methods: A total of 20 stroke survivors with PSS underwent pendulum tests with smartphones from multiple angles to quantitatively assess knee muscle spasticity. Pose estimation was conducted using the AlphaPose and STCFormer algorithms, with simultaneous measurements using an electronic goniometer as a reference. Three pendulum parameters were evaluated: normalized relaxation index (P1), first maximum of oscillation (P2), and relaxation index at half swing (P3). Bland–Altman analyses were used to analyze the consistency between pose estimation and electronic goniometer measurements. Intraclass correlation coefficient (ICC) and Spearman’s correlation analyses were conducted to evaluate agreement and reliability between electronic goniometer measurements and clinical evaluation. Results: P1 demonstrated the highest consistency between pose estimation and electronic goniometer measurements, with the highest ICC values (0.931 for AlphaPose and 0.911–0.94 for STCFormer). P1 and P3 differentiated between affected and unaffected limbs (p < 0.01) and demonstrated significant negative correlations with Modified Ashworth Scale scores, particularly for knee extensors (P1: ρ = −0.747 for AlphaPose and −0.781 for STCFormer; p < 0.01). P2 demonstrated low consistency and differential performance across all analyses. Conclusions: Video-based human pose estimation, particularly using P1, offers a reliable and objective method for evaluating PSS, demonstrating strong agreement with electronic goniometer measurements. This approach is clinically feasible for evaluating spasticity.