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: this paper does not state a specific Theia3D software version in its methods.
Study Overview
Design
A concurrent validation study comparing spatiotemporal gait measures between marker-based and markerless (Theia3D) motion capture during overground and treadmill walking in people with Parkinson's disease, examining both a custom gait-analysis approach and the markerless system's built-in automatic report.
Participants
Thirty individuals with Parkinson's disease (mean age 61.4 years), diagnosed by a specialist, assessed while on their regular Parkinson's medication, with a mean Movement Disorder Society-Unified Parkinson's Disease Rating Scale Part III score of 19.4, indicating mild motor symptom severity.
Data Collection
Participants walked along a 12-meter walkway 10 times at a comfortable pace, twice with a short break between, followed by two one-minute treadmill walking trials at their self-selected overground pace. For both conditions, gait was recorded separately using marker-based motion capture (three markers on each foot) and markerless motion capture processed through Theia3D, with the order of the two methods randomized and balanced according to disease stage and freezing-of-gait status.
Analysis
Markerless gait data were analyzed two ways: using a custom algorithm that detected gait events from virtual marker velocities, matching the approach used for the marker-based data, and separately using the markerless system's built-in automatic analysis report, which uses a different, pelvis-to-heel-based method for detecting gait events. Agreement between methods was assessed using intraclass correlation coefficients, Pearson correlation coefficients, mean differences, and Bland-Altman analysis, with disease severity additionally tested as a potential influence on the degree of agreement observed.
Key Findings
- When analyzed using the custom, population-appropriate gait-detection approach, markerless motion capture showed good to excellent agreement with marker-based motion capture for nearly every spatiotemporal gait measure during overground walking, with double support time as the sole exception showing poor agreement.
- During treadmill walking, every spatial and temporal gait measure showed excellent agreement between the two systems, suggesting markerless capture performs particularly well in this more controlled walking condition.
- Differences between the markerless and marker-based systems for gait speed and step length were smaller than commonly used thresholds for a practically meaningful change in this population, supporting markerless capture's ability to reliably detect genuine changes in these specific measures.
- Severity of Parkinson's disease motor symptoms, as measured by a standard movement rating scale, was not correlated with the degree of disagreement between the two systems, indicating that markerless accuracy for these measures was not meaningfully affected by how severely a person's movement was affected.
- In clear contrast, the markerless system's own built-in automatic analysis report showed poor agreement with marker-based measurement for most gait measures, with only gait speed and step length showing good agreement; the researchers attribute this gap specifically to differences in how gait events and measures are calculated within that built-in pipeline, rather than to any underlying limitation in the markerless body-tracking itself, since the same markerless tracking data produced considerably better agreement when analyzed with the custom, population-suited approach.
What This Means for Movement Science Researchers and Parkinson's Disease Assessment Teams
For researchers and assessment teams working with Parkinson's disease populations, this study offers an encouraging core finding with one specific, important caveat attached: Theia3D markerless motion capture can reliably measure most spatiotemporal gait parameters in this population, particularly during treadmill walking, when paired with a gait-analysis approach suited to Parkinsonian gait characteristics. The caution around the system's built-in automatic report is equally important and practically actionable: teams or staff without specialized movement-analysis expertise who rely on default, out-of-the-box analysis outputs in this population should be aware that those specific results showed meaningfully weaker agreement with marker-based measurement in this study, reinforcing the value of using, or developing, gait-event detection approaches validated specifically for movement differences like those seen in Parkinson's disease rather than defaulting to approaches built primarily around healthy, typical gait patterns.
The Role of Theia3D in Movement Analysis
Theia3D was the markerless motion capture system directly validated against marker-based motion capture in this study, specifically for its ability to measure spatiotemporal gait parameters, rather than full joint kinematics, in a population with Parkinson's disease. The study's additional comparison between a custom analysis approach and the system's built-in automatic report offers a useful, broader methodological insight: that measurement discrepancies in markerless motion capture research can originate from how collected data is processed and gait events are detected, not only from the underlying camera-based body tracking itself, a distinction relevant to interpreting markerless validation research more generally.
Study Limitations and Future Directions
The authors note that due to a limited number of available cameras, they could not record the marker-based and markerless systems simultaneously during overground walking, meaning the two methods could not be compared on the exact same individual walking trials for that condition, though the treadmill condition allowed them to confirm participants walked at matched speeds for both methods, partially addressing this limitation. They also note their custom gait-event detection algorithm, while previously validated and shown to be robust in people with movement-related neurological differences, may still carry its own limitations relative to the markerless system's built-in automatic report. The authors identify establishing an improved, fully automated analysis pipeline specifically suited to people with Parkinson's disease and other movement-related neurological differences, and extending this kind of validation to additional gait characteristics such as asymmetry and across different Parkinson's disease subtypes and disease stages, as important next steps for this line of research.
Citation: Seuthe, J., Barbieri, F. A., Grotherr, J., Hauptmann, B., & Schlenstedt, C. (2025). Validation of markerless motion capture for spatiotemporal gait measures in people with Parkinson's disease. Journal of Biomechanics, 193, 113008. https://doi.org/10.1016/j.jbiomech.2025.113008
Link: Read the full study
Does Markerless Capture Hold Up for Gait Assessment in Parkinson's Disease?
This study of 30 individuals with Parkinson's disease finds that Theia3D shows good to excellent agreement with marker-based motion capture for most spatiotemporal gait measures, when paired with an appropriate gait-detection approach, while cautioning that the system's built-in automatic report showed weaker agreement in this population. If your work involves movement assessment in Parkinson's disease or other movement-related neurological conditions, talk to our team about applying Theia3D to your own research.