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 v2022.1.0.2309, as stated in the abstract.
Study Overview
Design
A concurrent, within-subject comparison in which Theia3D markerless and a marker-based system (CAST model) recorded the same overground walking trials simultaneously, for both adults and children with diagnosed gait disorders.
Participants
Thirty-four subjects with orthopedic or neuro-orthopedic gait disorders: 10 adults aged 20 to 55 (4 men, 6 women) and 24 children and adolescents aged 7 to 17 (20 boys, 4 girls).
Data Collection
Eight Theia3D video cameras were synchronized with an existing 11-camera marker-based (Qualisys Oqus) system already in place in the lab. Subjects walked barefoot overground at a self-selected pace; marker-based kinematics used the CAST model. Each subject completed one static calibration trial and six walking trials, sampled at 50 Hz.
Analysis
Lower-body 3D kinematics were estimated separately from the Theia3D (v2022.1.0.2309) and marker-based (Visual3D) data for each gait cycle. Gait cycles were time-normalized to 101 points, and root mean square error (RMSE) and Pearson's correlation coefficient were calculated between the two systems for each joint angle, on both sides.
Key Findings
- Hip rotation showed the weakest agreement of any measure. Theia3D data showed a reduced range of motion and a continuous slight external rotation not seen in the marker-based data, with an RMSE of 9° and a correlation coefficient of 0, meaning the two systems' hip rotation curves did not track together at all.
- Knee rotation also disagreed substantially between systems. RMSE reached 9.85° with a correlation coefficient of just 0.14, placing it alongside hip rotation as the two clearest points of disagreement.
- Pelvic tilt, hip extension, and knee flexion showed smaller, consistent offsets rather than a breakdown in agreement. Theia3D pelvic tilt was posteriorly shifted (RMSE 6.32°), hip extension was increased (RMSE 6.87°), and knee flexion during terminal stance was slightly decreased (RMSE 3.85°), alongside frontal-plane differences including decreased downward pelvic obliquity, decreased hip abduction in initial swing, and increased knee valgus in the Theia3D data.
- These pathological-gait findings closely paralleled the authors' own earlier work in healthy adults. The authors note the pattern of agreement and disagreement reported here corroborates similar RMSE and correlation values from their companion study in healthy subjects (Schnell et al., 2022, Gait & Posture, 97, S332–S333), and that the lower spread observed in the Theia3D data here is consistent with findings reported elsewhere in the literature, though pelvis data were not available for comparison in that case.
What This Means for Movement Science Researchers
For researchers evaluating whether markerless capture holds up outside of healthy, typical gait, this study offers a cautiously encouraging answer: most lower-body kinematics showed good agreement even in a mixed pediatric and adult sample with diagnosed gait disorders, but hip and knee transverse-plane rotation remain the measures to treat with the most caution, consistent with patterns seen in other published Theia3D validation work. The authors themselves note that comparisons between marker-based models can differ from each other too, so disagreement with one specific marker-based reference doesn't necessarily mean Theia3D is wrong, just that further comparison work is warranted.
The Role of Theia3D in Movement Analysis
In this study, Theia3D provided a markerless alternative for capturing lower-body 3D kinematics during overground walking trials in a population, young children with mobility challenges and adults with gait disorders, where marker-based preparation can be especially burdensome. Running simultaneously with the lab's existing marker-based system let the researchers compare the two directly, gait cycle by gait cycle, across the same pathological walking trials.
Study Limitations and Future Directions
This summary is based on a brief conference abstract (ESMAC 2023 Abstract Book) rather than a full peer-reviewed journal article, so some methodological detail available in a full paper isn't available here: there's no breakdown of how orthopedic versus neuro-orthopedic diagnoses were distributed across the sample, and the reported results are largely descriptive (RMSE and correlation coefficients) without formal inferential statistics. The authors themselves note that differences exist between marker-based models, and call for further work comparing Theia3D against additional marker-based references, not just the one used here, before drawing firmer conclusions, particularly for hip and knee rotation.
Citation: D'Souza, S., Doepner, R., & Fohanno, V. (2023). Comparison of lower-body 3D-kinematics between Theia3D markerless and the CAST model marker-based systems during pathological gait in adults and children. Gait & Posture, 106, 40–41 (ESMAC 2023 Abstract Book).
Link: Read the full study
Does Markerless Validation Hold Up Beyond Healthy Gait?
Pathological gait rarely looks like the textbook stride most validation studies are built around. If your research involves adults or children with diagnosed gait disorders and you want to understand where markerless capture tracks closely and where it needs a second look, talk to our team about what this kind of validation work means for your specific population.