A model‐based tracking method for measuring 3D dynamic joint motion using an alternating biplane x‐ray imaging system

PURPOSES To propose a new model-based tracking method for measuring three-dimensional (3D) dynamic joint kinematics using a clinical alternating biplane x-ray imaging system; and to quantify in vitro its errors in measuring ankle and knee motions at different motion speeds. METHODS A new model-based tracking method based on motion component partition and interpolation (MCPI) was developed for measuring 3D dynamic joint kinematics based on a clinical alternating biplane x-ray imaging system. Two detectors of the biplane imaging system placed perpendicular to each other were operated to collect alternating fluoroscopic images of the targeted joint during tasks. The CT data of the joint were also acquired for the reconstruction of volumetric and surface models of each of the associated bones. The CT-based models of the bones were first registered to the alternating images using a model-to-single-plane fluoroscopic image registration method, and the resulting bone poses were then refined using a two-level optimization with motion component partition and model vertex trajectory interpolation. The MCPI method was evaluated in vitro for measurement errors for an ankle and a knee specimen moving at different speeds against a standard reference provided by a highly accurate motion capture system. The positional and rotational errors of the measured bone poses were quantified in terms of the bias, precision, and root-mean-squared errors (RMSE), as well as the mean target registration error (mTRE), a final mTRE less than 2.5 mm indicating a successful registration. RESULTS The new method was found to have RMSE of bone pose measurements of less than 0.18 mm for translations and 0.72° for rotations for the ankle, and 0.33 mm and 0.74° for the knee with a high successful registration rate (>97%), and did not appear to be affected by joint motion speeds. Given the same alternating fluoroscopic images, the MCPI method outperformed the typical biplane analysis method assuming zero time offset between the two fluoroscopic views. The differences in performance between the methods were increased with increased joint motion speed. With the accurate bone pose data, the new method enabled talocrural, subtalar, and tibiofemoral kinematics measurements with submillimeter and subdegree accuracy, except for an RMSE of 1.04° for the internal/external rotation of the talocrural joint. CONCLUSIONS A new model-based tracking method based on MCPI has been developed for measuring dynamic joint motions using an alternating biplane x-ray imaging system widely available in medical centers. The MCPI method has been demonstrated in vitro to be highly accurate in determining the 3D kinematics of the bones of both the ankle joint complex and the knee. The current results suggest that the MCPI method would be an effective approach for measuring in vivo 3D kinematics of dynamic joint motion in a clinical setting equipped with an alternating biplane x-ray imaging system.