Signal decomposition method of evaluating head movement to measure induced forelimb lameness in horses trotting on a treadmill.
Authors: Keegan K G, Pai P F, Wilson D A, Smith B K
Journal: Equine veterinary journal
Summary
# Editorial Summary Quantifying mild forelimb lameness in horses remains challenging because absolute measurements of vertical head displacement often lack sensitivity to subtle gait changes, whilst behavioural head movements from nervousness or curiosity introduce measurement noise that masks lameness-induced asymmetry. Keegan and colleagues developed a signal decomposition technique to isolate three components of vertical head movement at trot: the lameness-induced asymmetry (A1), the natural biphasic head bobbing pattern (A2), and extraneous behavioural movement, mathematically extracting the noise to leave a cleaner lameness signal. Testing this method on nine horses with experimentally induced mild forelimb lameness, A1 increased by a mean of 1.63 cm (range 0.10–3.33 cm) following lameness induction, whilst A2 remained unchanged, with excellent reproduction accuracy of 99.5–99.7 per cent. For practitioners and researchers using treadmill-based kinematic analysis, this approach offers substantially improved sensitivity for detecting early-stage or mild lameness that might otherwise escape detection, particularly valuable in clinical trials or when evaluating subtle performance-limiting conditions where traditional visual assessment proves unreliable.
Read the full abstract on PubMed
Practical Takeaways
- •Signal decomposition of vertical head movement provides a more sensitive method to detect mild forelimb lameness than absolute head height measurements alone
- •This computer-assisted kinematic technique filters out extraneous head movement from nervousness or excitement, improving measurement reliability in clinical trials
- •The method shows promise for objective lameness evaluation in treadmill-based clinical studies, requiring relatively modest sample sizes (n=12) for statistical power
Key Findings
- •Signal decomposition method successfully separated lameness-induced head movement (A1) from natural biphasic movement (A2) and extraneous movement with 0.3-0.5% error
- •Mean A1 increased by 1.63 cm (range 0.10-3.33 cm, P=0.005) following induced lameness
- •Mean A2 did not significantly change after lameness induction, confirming the decomposition model
- •Sample size calculation indicated 12 subjects needed for 80% power to detect lameness differences using this method