Visual estimation of radioulnar incongruence in dogs using three-dimensional image rendering: an in vitro study based...

To determine the sensitivity and specificity of visual estimation of radioulnar incongruence (RUI) in the canine elbow by use of 3-dimensional (3D) image rendering.

Experimentally induced negative and positive RUI.

Canine (>20 kg) cadaveric right thoracic limbs (n=8).

Radial shortening and lengthening of 1 and 2 mm were performed extending an established surgical in vitro model of RUI. Based on transverse computed tomographic (CT) scans of each radioulnar conformation, the subchondral radioulnar joint surface was reconstructed and visualized. A total of 64 3D models of RUI were blindly evaluated in a random manner by 3 independent observers. RUI was estimated subjectively at 1 mm precisely (-2, -1, 0, +1, +2) by visual inspection of the 3D models.

Median sensitivity for identifying an incongruent joint was 0.86. Median specificity for identification of a congruent joint was 0.77. Analyzing the data only in respect to a congruent joint versus one with a shortened radius (positive RUI) resulted in a median sensitivity of 0.82, and a median specificity of 1.00. Interobserver agreement was 0.87. Repeatability was 0.96.

Estimation of positive and negative RUI based on 3D surface models of the radioulnar articulation mimics gross inspection in a noninvasive manner, the latter being the ultimate gold standard for definitive diagnosis of any radioulnar step. The proposed technique is precise, reliable, and repeatable in vitro.

Preoperative estimation of the type and degree of RUI is the basis for deciding which type of corrective or modifying osteotomy might be best suited to restore normal joint loads in vivo.