Study at a glance
Summary
Disclaimer: The following summary details independent academic research. Theia3D is intended for research, sports performance and educational use only. It is not cleared as a medical device and is not intended to be used for the diagnosis, treatment, mitigation or prevention of any disease, injury or medical condition.
Editor's note: Theia3D version 2022.1.0.2309, as stated in the paper. Here, Theia3D served as the markerless motion-capture tool for a research question about the reliability of agility T-test biomechanics, rather than as a system being validated against marker-based motion capture within this particular study; the authors cite prior published Theia3D validation research to support their use of the system.
Study Overview
Design
A single-group, repeated-measures reliability (test-retest) study assessing the between-session consistency of performance, knee functionality, and movement quality biomechanical parameters during a side-shuffling agility T-test, using markerless motion capture.
Participants
Sample size was calculated using established methodological frameworks for reliability studies, targeting 18 recreationally healthy adults. One female participant sprained her ankle during warm-up in the second session and could not complete testing, leaving 17 participants (11 male, 6 female; age 24.4 ± 2.6 years; BMI 23.4 ± 2.1; 4 left-leg dominant, 13 right-leg dominant) in the final analysis. Individuals with neuromuscular or musculoskeletal impairments, lower-extremity trauma in the prior six months, or other conditions that could affect performance were excluded during recruitment.
Data Collection
Participants completed two test sessions exactly one week apart, tested at the same time of day to avoid circadian effects, following a separate familiarization session before the first test. Each session included six trials of the side-shuffling agility T-test (three trials shuffling first to the left, three to the right), with 3 to 5 minutes of rest between trials. Movement was captured using 10 synchronized Qualisys Miqus video cameras (1280 x 720 pixels, 180 Hz). Video data were processed using Theia3D to compute inverse-kinematic three-dimensional poses, with a kinematic chain configured to allow six degrees of freedom at the trunk-pelvis and three degrees of freedom at the hip, knee, and ankle joints; the resulting poses were imported into Visual3D to calculate time durations and peak acceleration/deceleration for each test phase (performance parameters), sagittal-plane knee joint angles, angular velocities, and whole-body deceleration during the initial forward deceleration phase (knee functionality parameters), and peak knee valgus motion, foot rotation, and contralateral pelvic tilt during the sideward shuffling phase (movement quality parameters).
Analysis
Test-retest reliability was evaluated using a two-way random-effects intraclass correlation coefficient (ICC (2,1), absolute agreement), interpreted as excellent (≥0.90), good (0.75-0.89), moderate (0.50-0.74), or poor (<0.50). Standard error of measurement (SEM) and minimal detectable change (MDC), which provides a threshold for distinguishing a true change in performance from measurement error, were calculated for all parameters.
Key Findings
- Performance timing measures showed good-to-excellent reliability: forward sprinting achieved an ICC of 0.88, while sideward (ICC 0.96) and backward (ICC 0.95) movement phases showed even higher reliability, with small measurement error (SEM ≤ 0.04 seconds) and minimal detectable change values below 0.12 seconds for phase durations.
- Peak acceleration and deceleration measures showed moderate-to-excellent reliability (ICC range 0.65 to 0.91); deceleration during the sideward shuffling phase showed the greatest variability, with a measurement error of 1.55 m/s² and a minimal detectable change of 4.3 m/s², compared with considerably lower values in the other phases.
- Knee functionality parameters showed good-to-excellent reliability overall: peak knee flexion (ICC 0.85) and extension angle (ICC 0.94) were both robustly reliable, sagittal knee angular velocities ranged from ICC 0.80 to 0.89, and whole-body deceleration at a standardized 50-degree knee flexion point showed good-to-excellent reliability (ICC 0.85 to 0.94).
- Among the movement quality parameters, peak knee valgus motion showed good-to-excellent reliability (ICC 0.83 to 0.90) with measurement error near 1.2 degrees, but transverse-plane foot rotation and frontal-plane contralateral pelvic tilt showed only moderate reliability (ICC 0.54 to 0.70), with comparatively larger measurement error (approximately 19% to 25% of the measured range) and minimal detectable change values, indicating more limited precision for detecting small individual-level changes in these two specific measures.
- The authors describe this as the first study to comprehensively quantify agility T-test movement biomechanics using markerless motion capture, and position the reported measurement error and minimal detectable change values as reference benchmarks to help interpret whether a future observed change in these parameters likely reflects a genuine change in performance rather than measurement variability.
What This Means for Movement Science Researchers and Return-to-Sport Testing Performance Teams
For researchers and performance staff using the agility T-test as part of return-to-sport or general athletic screening, this study offers a specific, practically useful foundation: rather than relying solely on overall completion time, markerless motion capture can reliably add detailed knee functionality and most movement quality metrics to this already widely used field test, broadening what a single testing session can reveal about an athlete's movement patterns. The study's reference measurement error and minimal detectable change values give practitioners concrete benchmarks for judging whether a change observed across testing sessions, say, after a period of targeted training, likely reflects a real change rather than typical test-retest variability, while the comparatively weaker reliability found for foot rotation and pelvic tilt is a useful, specific caution against over-interpreting small session-to-session differences in those two particular measures.
The Role of Theia3D in Movement Analysis
Theia3D served as the markerless motion-capture tool in this study, enabling detailed three-dimensional kinematic measurement of a fast, multidirectional field-based test that would be considerably more logistically demanding to assess repeatedly with a marker-based system confined to a specialized laboratory. The authors cite prior published Theia3D validation research reporting sagittal-plane joint-angle agreement with marker-based motion capture of approximately 3° to 6° for the ankle and knee, with somewhat larger hip-angle differences (roughly 7° to 16°) during high-intensity tasks like running and cutting, to support confidence in applying the system to this multi-planar, high-demand agility task.
Study Limitations and Future Directions
The authors note several limitations specific to this study. One participant's ankle sprain during warm-up in the second session reduced the final analyzed sample from the targeted 18 to 17 participants, which the authors note likely widened the confidence intervals for some reliability estimates, particularly for parameters that already showed comparatively lower reliability; some estimates should therefore be interpreted with a degree of caution given the resulting statistical uncertainty. The study's young, physically active sample may not generalize to older adults or less physically active populations. While knee functionality parameters showed moderate-to-excellent reliability overall, some differences were observed between dominant and non-dominant legs, and the authors note that no clear, well-established threshold currently exists for distinguishing typical from atypical between-limb asymmetry. Finally, this study included only healthy participants and did not include individuals managing an existing lower-limb joint concern, such as those recovering from ACL reconstruction, whose movement variability may differ due to compensatory movement patterns in the affected limb; the side-shuffling task's double-limb-support design, unlike the single-leg plant-and-cut movement typical of many change-of-direction tasks, and the absence of force-plate data to confirm movement-event timing, may also have contributed to the comparatively weaker reliability observed for the foot rotation and pelvic tilt measures specifically.
Citation: Wen, Z., Claeys, K., González García, J., Jamkrajang, P., Xu, J., Luyckx, T., & Vanrenterghem, J. (2026). Test-retest reliability of performance, knee functionality and movement quality biomechanics in agility T-test using markerless motion capture. Journal of Electromyography and Kinesiology, 89, 103165. https://doi.org/10.1016/j.jelekin.2026.103165
Link: Read the full study
How Reliable Is Markerless Motion Capture for Measuring Agility T-Test Biomechanics?
This study of 17 healthy adults uses Theia3D markerless motion capture to show moderate-to-excellent between-session reliability for performance timing and knee functionality parameters during the side-shuffling agility T-test, with slightly weaker reliability for foot rotation and pelvic tilt measurements, establishing reference benchmarks for future field-based agility testing. If your work involves return-to-sport screening or agility performance research, talk to our team about applying Theia3D to your own studies.