Abstract| Volume 29, SUPPLEMENT 1, S317-S318, April 2021

Test-restest repeatability and reproducibility of 3D joint space mapping at the ankle using weight bearing cone beam computed tomography

      Purpose: Joint space mapping (JSM) is an image analysis tool that can measure, display, and set up for analysis of joint space width (JSW) distribution in 3D. It has been applied successfully at the hip and knee and is now demonstrated at the ankle using weight bearing cone beam computed tomography (CBCT). Our objective was to demonstrate: (1) feasibility of JSM at this site of complex anatomy in the presence of metalwork; (2) the effect of talocrural joint angulation on 3D JSW distribution; (3) JSM test-retest repeatability; and (4) JSM interoperator reproducibility.
      Methods: A convenience sample of 25 individuals with repeat weight bearing CBCT imaging of both feet and ankles performed within 4 months between 2013 and 2017 were retrospectively selected with no prior constraints on joint positioning. Imaging was acquired with a Curvebeam pedCAT scanner, 120 kVp, 0.37 mm isotropic voxels, FOV diameter 35 cm x height 20 cm, with a sharp reconstruction kernel. 2 individuals with metal-induced artefact were excluded because of failure of the JSM measurement algorithm.
      The mean ± sd age of the 23 study individuals was 52.7 ± 14.7 years, with 16 females and 7 males. Sides for analysis were selected to provide a mixture of pathology (13 left, 10 right). 11 ankles had no metalwork, 6 various metalwork fusions in the ipsilateral foot, 6 in the contralateral foot. The mean ± sd interval between imaging visits was 74.0 ± 29.6 days.
      JSM was performed by a single blinded operator at the medial talocrural (MTC, blue patch in first figure), lateral talocrural (LTC, purple), talonavicular (TN, red), and posterior subtalar (PST, orange) articular surfaces at both visits.
      A second operator trained on 6 opposite ankles, then performed JSM patch segmentation at each of the 23 baseline ankles. All results were registered to a set of average joint surfaces.
      Talocrural joint angulation was measured by a single blinded operator at each ankle from 3D reconstructions as the angle set by the lines connecting landmarks at the centre of the distal tibial diaphysis, the centre of the talar dome, and the centre of the talar head articular surface. The paired t-test was used to determine whether mean angle difference between visits significantly differed from 0°. Dependence of 3D JSW on angle difference and the null hypothesis of no difference in 3D JSW between visits were tested using statistical parametric mapping (SPM).
      Baseline and follow-up maps were used to show test-retest repeatability, while comparison of baseline maps was used to show interoperator reproducibility, both as Bland-Altman statistics.
      Results: Mean JSW values at visit 1 are shown in 3D on the average set of joint surfaces (second figure).
      The difference in angulation between visits was not significantly different from zero (2.7 ± 17.7°, p=0.30). SPM showed no significant dependence of difference between baseline and follow up 3D JSW on joint angulation nor any significant difference in 3D JSW difference from zero for between visits (threshold p<0.05).
      Global bias was 0.0 mm across all surfaces for test-retest repeatability and reproducibility. Repeatability limits of agreement at individual surfaces were: TN ±0.26 mm, MTC ±0.28 mm, ST ±0.34 mm, and LTC ±0.40 mm. Optimum sensitivity was less than ±0.2 mm at each central joint surface. Reproducibility limits of agreement at individual surfaces were: TN ±0.27 mm, MTC ±0.24 mm, ST ±0.31 mm, and LTC ±0.37 mm. 3D maps show the distribution of repeatability (third figure) and reproducibility (fourth figure) metrics across each joint surface.
      Conclusions: This is the first report of 3D surface-based measurement and analysis of JSW at this particular set of complex articular surfaces. JSM is feasible, reproducible, and sensitive in 3D JSW measurement from weight bearing CBCT to at least ±0.2mm across the ankle articular surfaces, which is subvoxel performance at nearly half of the isotropic voxel dimension. Uncontrolled angulation of the talocrural joint also appears not to be significantly different between baseline and follow-up imaging nor have any significant effect on 3D JSW distribution. Although there is a limit, if tolerated then JSM can still be performed in the presence of metalwork, which is a common feature for this particular patient population.