Biosystems One / Engineering verification example
MotionLab synthetic tracking validation note
Document version: 1.0 Date: August 7, 2026
Scope: This is an example engineering-verification report based on a deterministic synthetic fixture. It is not a human-subject study, clinical validation, protocol validation, or general accuracy claim.
Intended test
The fixture evaluates whether the two-dimensional point-tracking implementation can follow a known cyclic trajectory under controlled synthetic conditions while preserving frame-level provenance.
Method
- Known ground-truth coordinates are generated before tracking.
- The tracker is initialized from a manual seed.
- Estimated coordinates are compared with ground truth by frame.
- Occlusion, re-acquisition, consistency, and segment-constraint paths are exercised elsewhere in the deterministic suite.
Recorded result
| Fixture | Completion | RMS position error | Interpretation |
|---|---|---|---|
| Synthetic cyclic trajectory | 36 / 36 frames | 0.306 px | Regression result for this fixture only |
Limitations
Synthetic performance does not establish accuracy on arbitrary cameras, codecs, subjects, tasks, marker conditions, occlusion patterns, or protocols. Large displacement, blur, rotation, scale change, deformation, and out-of-plane movement can require correction. AI-assisted landmarks remain estimates and require comparison with independently digitized recordings for protocol-specific research use.
Recommended external validation
For an intended protocol, use independently digitized real recordings and report RMS error, 95th-percentile error, lost-track rate, false-lock rate, and inter-rater variability.