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 2025.1.2, as stated in the paper. This is one of the more directly critical papers regarding Theia3D in this collection, and the summary above and findings below are presented without softening, consistent with our practice of flagging author-identified limitations clearly rather than minimizing them.
Study Overview
Design
A retrospective, cross-sectional concurrent-validity study comparing ankle kinematics from marker-based and markerless (Theia3D) motion capture in children with cerebral palsy wearing ankle-foot orthoses, stratified by orthosis type into a sagittal motion-permitting group and a sagittal motion-restricting group, using each orthosis type's known mechanical constraints as an additional reference point for interpreting agreement between systems.
Participants
Forty children with cerebral palsy (community ambulators, Gross Motor Function Classification System levels I-III) who wore AFOs for everyday mobility and underwent gait assessment at a pediatric movement-assessment center over a three-year period. Participants were grouped by orthosis design: 29 participants (36 limbs) wore sagittal motion-permitting AFOs (hinged or posterior leaf-spring), and 11 participants (19 limbs) wore sagittal motion-restricting AFOs (solid or ground-reaction).
Data Collection
Participants walked at a self-selected pace across a 10-meter walkway while wearing shoes over their AFOs. Reflective markers were placed on the shoe or orthosis itself, since the orthoses limited direct access to underlying foot and ankle landmarks. Marker-based and markerless data were captured simultaneously using a synchronized Qualisys system of 10 marker-based cameras and 10 markerless Miqus cameras.
Analysis
Ankle kinematics were compared between systems across the full walking cycle using statistical parametric mapping, and the magnitude of between-system differences was quantified using root mean square deviation (RMSD), with values above 5 degrees considered a commonly used threshold for a practically meaningful difference. Range of motion was compared between systems using paired statistical tests, separately for each AFO group.
Key Findings
- Both AFO groups showed statistically significant differences between marker-based and markerless ankle kinematics across all three planes of motion (sagittal, frontal, and transverse).
- Markerless-derived sagittal-plane range of motion significantly exceeded marker-based range of motion in both groups (motion-permitting AFOs: 19.3° markerless vs. 14.5° marker-based; motion-restricting AFOs: 16.0° markerless vs. 9.1° marker-based).
- The transverse plane showed the largest and most practically concerning discrepancy: markerless range of motion was roughly three times greater than marker-based range of motion in both groups (motion-permitting AFOs: 18.3° vs. 6.1°; motion-restricting AFOs: 17.1° vs. 5.3°), with markerless values considerably exceeding what the orthoses' physical design is intended to allow.
- In the frontal plane, the two systems differed by a consistent offset throughout the walking cycle, with markerless output shifted toward greater inversion, but overall range of motion was comparable between systems in both groups, suggesting a systematic positional offset rather than a tracking disagreement in this specific plane.
- Average RMSD between systems was near or below the 5-degree threshold in the sagittal plane for both groups but exceeded it in the frontal and transverse planes for both groups, with the transverse plane showing the largest discrepancies (8.98° to 9.65°).
- The authors note a similar pattern of exaggerated transverse-plane ankle motion has also appeared in prior Theia3D research involving children and adults not wearing orthoses, including in the company's own published validation materials, pointing toward a broader characteristic of how the markerless algorithm defines the foot segment rather than an issue specific to AFO use or this population.
What This Means for Movement Science Researchers and Pediatric Gait Assessment Teams
For researchers and assessment teams working with children with cerebral palsy, and particularly those using ankle-foot orthoses, this study offers a clear, specific caution: markerless ankle kinematics in this population, especially in the transverse plane, should currently be interpreted with real caution rather than treated as directly interchangeable with marker-based measurement. The authors are explicit that these discrepancies exceed what the orthoses' own physical design should allow, and they frame this as a meaningful, addressable limitation in how the markerless algorithm models the foot and ankle segment under constrained conditions, rather than evidence against markerless capture's broader usefulness. Teams whose work depends on precise ankle angle values in braced or similarly constrained populations, particularly to inform movement-related decisions, should weigh this finding carefully and consider confirming markerless ankle findings against marker-based measurement, or against device-specific mechanical plausibility checks like the one used in this study, before relying on markerless ankle kinematics alone in this specific population and measurement combination.
The Role of Theia3D in Movement Analysis
Theia3D was the markerless motion capture system directly evaluated against marker-based motion capture in this study, specifically for ankle kinematics in children with cerebral palsy wearing ankle-foot orthoses. The study's use of the orthoses' known mechanical constraints as an independent plausibility check, alongside the standard marker-based comparison, offers a notably rigorous validation approach: rather than relying solely on agreement between two imperfect systems, the authors were able to assess whether the markerless system's output was physically possible given the device's own design, a useful methodological contribution for markerless validation research in populations using assistive devices with known mechanical properties.
Study Limitations and Future Directions
The authors note that their findings reflect a specific Theia3D software version (2025.1.2), and that results may not fully generalize to future versions, though they observe that the pattern found here is consistent with findings reported using earlier versions in related research. They note that the markerless system's underlying training data is unlikely to include individuals wearing reflective markers, and that the potential influence of marker presence on markerless landmark detection remains unknown. Because both limbs were included for participants wearing bilateral orthoses, some observations came from the same individual, which may have introduced a degree of non-independence not accounted for in the analysis. The authors also note that because the foot is enclosed within the shoe and orthosis for both systems, neither marker-based nor markerless motion capture measures true underlying skeletal foot motion directly, a limitation that likely affects both systems similarly. Orthosis stiffness, an important determinant of how much an orthosis actually constrains motion, was not directly measured and is inconsistently reported in the broader literature on these devices; grouping orthoses into just two categories was a practical simplification that may obscure more detailed relationships between orthosis properties and kinematic findings. The authors also note that AFOs themselves exhibit some compliance under load, meaning real motion can occur within the orthosis itself, so they should not be treated as a perfectly rigid reference standard. Finally, this study focused specifically on ankle kinematics during walking; the authors identify extending this kind of validation to other movement tasks, to force-related measures, and to other populations using assistive devices as important next steps.
Citation: Horn, R. J., Collins, J. D., Waters, H., Amacker, C., Amaechi, M., Thompson, R. M., & Curran, P. F. (2027). Validity of markerless motion capture for ankle kinematics in children with cerebral palsy wearing ankle-foot orthoses. Gait & Posture, 132, 110675. https://doi.org/10.1016/j.gaitpost.2026.110675
Link: Read the full study
Does Markerless Capture Accurately Measure Ankle Motion in Children Wearing Ankle-Foot Orthoses?
This study of 40 children with cerebral palsy finds that Theia3D shows significant differences from marker-based motion capture across all three planes of ankle motion, with markerless range of motion in the transverse plane substantially exceeding what the orthoses' own mechanical design should physically allow. If your work involves movement assessment in children using assistive devices, talk to our team about where Theia3D currently stands for this kind of application.