Quantitative Gait Analysis in a Sheep Model of In-Utero Spina Bifida Repair

Abstract

Objective: To evaluate quantitative gait analysis in a sheep model (Ovis aries) of in-utero spina bifida repair. Method: Using a 3D marker-based motion capture system, we quantified spatiotemporal gait parameters and hindlimb joint angles in a sheep model of in-utero spina bifida repair and normal controls. A linear support vector machine classifier, trained and tested on subject-disjoint partitions with individual gait cycles as the unit of analysis, evaluated the relationship between hindlimb joint angles and clinician-assigned Texas Spinal Cord Injury Scale scores. Linear mixed-effects models compared spatiotemporal parameters, hoof height, and hip drop across 3- to 6-month evaluations and between groups. Results: In treated (n=14) and control (n=3) lambs, front- and hindlimb gait speed differed by timepoint (both p=0.025). Across 1,632 left and 1,661 right hindlimb gait cycles, held-out classification accuracy was 85.9% and 88.6% (Cohen's {kappa}=0.78 and 0.84) for left and right hindlimbs, respectively, against majority-class proportions of 76.5% and 75%. Left hindlimb hip drop differed between treated lambs and controls ({Delta}=0.8 cm, p=0.027). Between-group contrasts were considered exploratory given the small control group. Conclusion: Quantitative gait analysis was feasible in an in-utero spina bifida repair model. Hindlimb kinematic profiles classified clinical ambulation scores above the majority-class baseline.