Comparing Inertial Measurement Units to Markerless Video Analysis for Movement Symmetry in Quarter Horses.
Authors: Pfau, Landsbergen, Davis, Kenny, Kernot, Rochard, Porte-Proust, Sparks, Takahashi, Toth, Scott
Journal: Sensors (Basel, Switzerland)
Summary
# Editorial Summary Quantifying lameness through movement analysis is becoming routine in equine practice, yet different measurement systems—from smartphone apps to wearable sensors—may yield varying results, potentially confounding clinical decision-making across referral centres and multi-site research. Pfau and colleagues evaluated whether a markerless artificial intelligence video analysis system could reliably detect the same movement asymmetries as validated inertial measurement units (IMUs) in 22 reining Quarter horses, comparing head and pelvic symmetry during in-hand trotting on contrasting surfaces and lunging work. Head movement asymmetry showed considerably wider limits of agreement between the two systems (29–50% in-hand, 22–38% on lunge), whilst pelvic movement asymmetry demonstrated substantially tighter concordance (13–24% in-hand, 14–24% on lunge)—importantly, these pelvic measurements fell within acceptable clinical thresholds for lameness identification. The tighter agreement for pelvic data is particularly valuable for practitioners relying on hind limb lameness assessment, which notoriously challenges subjective "eye" grading; however, the wider discrepancies in head movement suggest caution when using video-based systems alone for subtle forelimb detection, especially in multi-centre contexts where consistency across platforms is critical to diagnostic reliability.
Read the full abstract on PubMed
Practical Takeaways
- •Video analysis apps can be used alongside IMU systems in clinical settings, though pelvic measurements are more reliable than head measurements for lameness detection
- •Hind limb lameness assessment benefits particularly from these systems, as pelvic symmetry measurement is challenging to evaluate subjectively by eye alone
- •These tools are practical for referral centres and multi-site studies where standardized gait analysis is needed without expensive laboratory infrastructure
Key Findings
- •Proportional limits of agreement were wider for head movement (29-50% in-hand; 22-38% on lunge) than pelvic movement (13-24% in-hand; 14-24% on lunge) between IMU and video analysis systems
- •Between-system agreement limits exceed current lameness thresholds for identifying affected limbs but are similar to non-lame repeatability values
- •Pelvic movement asymmetry measurements showed narrower agreement limits, particularly valuable for detecting hind limb lameness
- •Video analysis and IMU systems demonstrate sufficient agreement for multi-centre studies and referral-level veterinary practice despite non-laboratory conditions