Clinical value of three-dimensional ultrasonography in the morphologic evaluation of rotator cuff tear: a prospective study

Objective To study the clinical value of three-dimensional ultrasonography (3D-US) in the morphological evaluation of rotator cuff tears (RCTs). Methods Based on previously published literature, RCT patterns in our study were divided into crescent, L-shaped with the remnant tendon retracted to the anterior rotator cuff (aL-shaped), L-shaped with the remnant tendon retracted to the posterior rotator cuff (pL-shaped), T-shaped (a tear pattern that is a combination of aL-shaped and pL-shaped), U-shaped, and massive type. Two radiologists prospectively assessed the tear patterns using 3D-US as well as magnetic resonance imaging (MRI) and compared these results using arthroscopy to calculate diagnostic accuracy. Result Fifty-two patients (52 shoulders) were enrolled. The overall diagnostic accuracy of 3D-US in evaluating RCT patterns (82.7%, 43/52; 95% CI: 72.1-93.3%) was significantly higher (p = 0.008) than that of the MRI (57.7%, 31/52; 95% CI: 45.8-73.4%). The accuracy of 3D-US was higher than that of MRI for most types of tears (crescent: 95.0% vs. 55.0%, aL-shaped: 83.3% vs. 77.8%, pL-shaped: 50.0% vs. 25.0%, T-shaped: 75.0% vs. 0.0%, and massive type: 80.0% vs. 100.0%). The accuracies of 3D-US with respect to evaluation by the two radiologists were 84.6% (44/52) and 76.9% (40/52), and there was substantial agreement evident (kappa = 0.709). The time taken by the two radiologists to reconstruct the 3D-US images and evaluate the tear pattern was < 5 min. Conclusion The 3D-US can be used for the preoperative evaluation of RCT patterns, and thus be useful for the correct selection of the surgical repair technique for RCTs.

Automated summary

The study demonstrates that the diagnostic accuracy of three-dimensional ultrasonography in the morphological evaluation of rotator cuff tears is higher than that of magnetic resonance imaging, especially for different types of tears. The two radiologists showed high accuracy in evaluating 3D-US images, with a short time required for reconstructing images and evaluating tear patterns.