Accuracy and Precision of Model-Based Tracking of a Dynamic Hop Landing Activity

TL;DR

This study evaluates the accuracy of a 3D imaging system for tracking knee movements during a hop landing activity, showing it is reliable for clinical research.

Contribution

The study provides new accuracy and precision metrics for BVR system use in dynamic hop landing simulations.

Findings

Kinematic accuracy ranged from 0.30 to 0.39° for rotations and 0.34 to 0.50 mm for translations.

Accuracy was consistent regardless of soft tissue presence or movement velocity.

The method is suitable for capturing clinically relevant differences in ACL-reconstructed patients.

Abstract

Biplane videoradiography (BVR) is the preferred 3D imaging modality for investigating the relationship between sub-millimeter knee kinematic abnormalities and posttraumatic osteoarthritis risk following anterior cruciate ligament (ACL) injury and surgery. Activity-specific BVR system geometries maximize BVR’s limited field of view which, in turn, influences downstream accuracy. The present work aimed to quantify the accuracy, bias, and precision of the reconstructed 3D tibiofemoral kinematics within a BVR system configured to capture the landing phase of a one-leg hop-for-distance activity. Radio-opaque beads were implanted into the femurs and tibiae of three cadaveric knees to provide the gold-standard kinematics. The specimens were moved through the BVR field of view simulating hop and drop landing motions such that the motion trajectories could better approximate dynamic in vivo velocities. The motions were tracked using both marker- and model-based methods. The mean absolute difference in kinematics between the two tracking methods was used to describe accuracy. Bland–Altman tests were used to quantify bias and precision. Kinematic accuracy ranged from 0.30 to 0.39° for rotations and from 0.34 to 0.50 mm for translations. The magnitudes of absolute difference, bias, and precision were similar regardless of the amount of soft tissue present or velocity of the simulated movement. Our results indicate that our approach for capturing BVR-derived kinematics for a one-leg hop-for-distance is sufficiently accurate to capture the magnitude of differences we expect to observe in a clinical population of ACL-reconstructed patients at long-term follow-up and will be useful to other investigators who may wish to record the hop-for-distance activity using the system geometry and image capture settings described here.

Figures