Biplane videoradiography with model-based tracking demonstrated mean angular RMS error of 1.37 ± 0.23° and mean translational RMS error of 2.05 ± 0.58 mm across foot and ankle joints, indicating accuracy suitable for in-vivo pathological research.
Key Findings
Results
The mean angular model-based tracking RMS error across all foot and ankle joints was 1.37 ± 0.23°.
Validation was performed against gold-standard radiostereometric analysis (RSA) using tantalum beads implanted in six bones: tibia, fibula, talus, calcaneus, cuboid, and navicular.
The greatest angular error was 1.66 ± 0.19° in the talonavicular joint.
The smallest angular error was 0.97 ± 0.19° in the subtalar joint.
Dynamic imaging was acquired during walking, step-up, step-down, and heel rise activities.
One participant was used for this validation study.
Results
The mean translational model-based tracking RMS error across all foot and ankle joints was 2.05 ± 0.58 mm.
The greatest translational error was 2.83 ± 0.21 mm in the calcaneocuboid joint.
The smallest translational error was 1.20 ± 0.29 mm in the subtalar joint.
Translational errors were compared against RSA-derived bone positions as the gold standard.
Error magnitude varied substantially across joints, with a range of over 1.6 mm between the best and worst performing joints.
Results
Joint accuracy errors exceeded bone accuracy errors, reflecting the compounded tracking of multiple bones.
Each joint's kinematics depends on the tracking accuracy of two separate bones, causing errors to compound.
The authors noted that 'joint accuracy errors are more clinically relevant and exceeded bone accuracy errors, reflecting compounded tracking of multiple bones.'
This compounding effect was identified as an important methodological consideration for interpreting BVR joint-level results.
The finding highlights that bone-level validation metrics alone may underestimate error at the joint level.
Results
Intra-rater repeatability as measured by coefficient of multiple correlations ranged from 0.545 to 0.996 across joints and tasks.
Coefficient of multiple correlation (CMC) analyses were used to quantify both inter- and intra-rater variability.
The range of 0.545 to 0.996 indicates substantial variability in repeatability depending on the specific joint and motion being assessed.
Both inter- and intra-rater variability were assessed as part of the validation pipeline.
Root mean square error and CMC analyses together were used to quantify accuracy and variability.
Methods
The BVR system was validated for six foot and ankle bones across four dynamic functional tasks.
Bones instrumented with tantalum beads included the tibia, fibula, talus, calcaneus, cuboid, and navicular.
Functional tasks assessed were walking, step-up, step-down, and heel rise.
This represents a comprehensive multi-bone, multi-task validation not previously reported for this specific BVR configuration.
The validation follows established practice requiring BVR validation against RSA for each new joint of interest and BVR configuration.
Six degrees of freedom (6DOF) kinematics were assessed for each bone using model-based tracking.
Conclusions
The authors concluded that BVR model-based tracking is accurate for assessing foot and ankle kinematics and can be translated to pathological research.
Pathological ankle conditions alter joint kinematics and contribute to degeneration in adjacent hindfoot and midfoot joints, motivating the need for accurate in-vivo assessment.
The validated pipeline is intended for use in understanding disease progression and guiding clinical interventions.
The study establishes a foundation for applying this BVR configuration to populations with conditions such as ankle osteoarthritis or other hindfoot and midfoot pathologies.
The authors noted that accurate in-vivo assessment of these motions 'is essential for understanding disease progression and guiding clinical interventions.'
What This Means
This research suggests that a specialized X-ray imaging technique called biplane videoradiography (BVR) can accurately measure how the small bones of the foot and ankle move during activities like walking, stepping up and down, and rising onto the heels. The technique works by tracking 3D bone movements in real time using two synchronized X-ray cameras, and the researchers validated its accuracy by comparing it to a gold-standard method called radiostereometric analysis (RSA), which uses tiny metal beads surgically implanted into bones as reference markers. One participant had tantalum beads placed in six bones of the foot and ankle, allowing a direct comparison between the two methods.
The results showed that the average angular measurement error was about 1.37 degrees and the average positional error was about 2.05 millimeters across all the foot and ankle joints studied. Accuracy varied by joint — for example, the subtalar joint (between the heel bone and the ankle bone) had the smallest errors, while the talonavicular and calcaneocuboid joints had slightly larger errors. The researchers also found that measuring how two bones move relative to each other (joint-level measurement) introduced more error than measuring a single bone alone, because small errors in tracking each bone add together.
This validation matters because conditions like ankle arthritis can alter how the small bones of the foot and ankle move, and these changes can spread to neighboring joints over time. Having a validated, non-invasive way to measure these subtle 3D motions during real activities could help researchers better understand how diseases progress and how treatments — like surgery or orthotics — actually change foot mechanics. This research suggests the BVR pipeline is now ready to be applied to patients with foot and ankle pathologies.
Check Your Own Numbers
Upload your bloodwork. We'll cross-reference your results against this study and 4,700 others.
Chauvet R, Bangsboll E, Bugajski T, Arndt A, Edwards W, Roach K. (2026). In-vivo validation of a high-speed biplane videoradiography pipeline for foot and ankle joint kinematics.. Journal of biomechanics. https://doi.org/10.1016/j.jbiomech.2026.113541