Skip to main content
2024 BMES Annual Meeting Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.

Biomechanics

Biomechanics - Poster Session E

Poster D7 - Reliability of Wearable Sensors to Measure Knee Angles During an ACL-IPP Screening Test

Saturday, October 26, 2024
10:00 AM - 11:00 AM EST
Location: Exhibit Hall E, F & G

Introduction:  Motion-capture (MC) technology, while limited to laboratory settings, is considered the gold-standard in collecting joint angle data [1]. However, recent studies have shown some success in obtaining reliable joint angle data using wearable sensors [2] which may enable relevant studies to take place outside of the lab. The purpose of this study was to quantify the accuracy of wearable sensor data collection in the context of an ACL-Injury Prevention Program (ACL-IPP) functional screening test application using a Bland-Altman analysis.

Materials and

Methods:  This project was approved by the Institutional Review Board at the University of Michigan (HUM00212750). Data from seven amateur female soccer players was collected across eleven data collection sessions. Knee angles were obtained using a 12-camera system with 43 markers and APDM wearable sensors (IMUs) on the thigh and shank. Participants performed five tasks (cutting, lateral shuffle, deceleration, drop-jump, triple-jump ) three times per leg. Knee angles were obtained using Visual3D and Moveo Explorer. Trials with IMU data exceeding normal knee motion ranges (external rotation beyond 45° [3], internal rotation beyond 25° [3], or flexion less than 0° or beyond 120° [4]) were excluded as outliers [Table 2]. The angle difference was calculated as MC minus IMU recorded angles. Shaded line graphs containing the mean angle difference with a 99% confidence interval were constructed for the overlapping time domain about maximum knee flexion [Fig 1]. Bland-Altman plots were constructed utilizing maximum knee flexion angles using Microsoft Excel to obtain the bias and 95% limits of agreement (LoA) [Table 1].

Results, Conclusions, and Discussions: Results
Data utilizing the right leg was used for analysis with positive angles assigned to flexion, adduction, and internal rotation angles. Most actions exhibited a bias of 5° or less for rotational motion. Adversely, nearly all actions exhibited abduction/adduction biases of -10° or greater while flexion/extension motion exhibited biases from 4° to 16° [Table 1]. Abduction angle differences exhibited a negative trend throughout time while the flexion angles displayed a positive trend [Fig 1]. Notably, the inlier rate for deceleration was 78%, indicating a sizeable number physiologically impossible readings were generated by the IMUs [Table 2]. The calculated 95% LoA for abduction ranged from a lower LoA of -50.01 (deceleration) to an upper LoA of 19.61 (lateral shuffle) [Table 1].

Discussion
This study showed mixed results in agreement between IMU sensors and motion-capture. While biases for rotational angles remained close to zero, with the IMUs tending to report rotational angles as around 5° more externally rotated (positive bias), there were some noticeable biases in flexion and abduction. For example, the IMUs tended to estimate abduction angles as less abducted/more adducted than the MC system (negative bias) whereas IMUs tended to record a smaller flexion value than the MC system (positive bias). The IMU sensors also reported large variability in abduction as indicated by a large 95% LoA range as well as consistent negative biases at max flexion across actions. It is also important to note that the IMU sensors generated sizeable amounts of physiologically impossible readings which could be derived from calibration issues or internal sensor fusion algorithm issues.

Conclusion
While there is some evidence that wearable sensors could be used in place of motion-capture systems in this study, IMU sensors were not able to accurately estimate knee abduction angles and showed mixed agreement for flexion across tasks. More research should be performed comparing IMU sensors to motion-capture systems to explore its current applications and limitations. Per the results of this study, current wearable sensor technology has the potential to be improved upon to increase reliability and accuracy of joint angle estimation when examining dynamic activities.

Acknowledgements (Optional):  References
[1] L. Wade, L. Needham, P. McGuigan, and J. Bilzon, “Applications and limitations of current markerless motion capture methods for clinical gait biomechanics,” PeerJ, vol. 10, p. e12995, Feb. 2022, doi: 10.7717/peerj.12995.
[2] D. Meldrum, C. Shouldice, R. Conroy, K. Jones, and M. Forward, “Test–retest reliability of three dimensional gait analysis: Including a novel approach to visualising agreement of gait cycle waveforms with Bland and Altman plots,” Gait Posture, vol. 39, no. 1, pp. 265–271, Jan. 2014, doi: 10.1016/j.gaitpost.2013.07.130.
[3] B. Zarins, C. R. Rowe, B. A. Harris, and M. P. Watkins, “Rotational motion of the knee*,” Am. J. Sports Med., vol. 11, no. 3, pp. 152–156, May 1983, doi: 10.1177/036354658301100308.
[4] G. Li, S. Zayontz, L. E. DeFrate, E. Most, J. F. Suggs, and H. E. Rubash, “Kinematics of the knee at high flexion angles: An in vitro investigation,” J. Orthop. Res., vol. 22, no. 1, pp. 90–95, Jan. 2004, doi: 10.1016/S0736-0266(03)00118-9.