MR imaging of rotator cuff tendon tears: comparison of T2*-weighted gradient-echo and conventional dual-echo...

Although MR imaging evaluation of the rotator cuff is usually done with proton density- and T2-weighted spin-echo techniques, interest also exists in T2*-weighted gradient-recalled echo techniques. The shorter scan times of T2*-weighted sequences can be used to increase the number of signal averages and thus improve the signal-to-noise ratio. Our purpose in this study was to compare the sensitivity and specificity of oblique coronal T2*-weighted MR image interpretations with conventional dual-echo T2-weighted MR image interpretations when diagnosing rotator cuff tears.

Forty-seven consecutive patients who underwent both MR imaging and shoulder arthroscopy were included in this study. The MR examination included both a 7 min 20 sec acquisition time proton density- and T2-weighted oblique coronal sequence and a 7 min 5 sec T2*-weighted oblique coronal sequence. The oblique coronal dual-echo T2-weighted images were evaluated by two of the authors independently, who were unaware of the arthroscopic findings. Each of the two observers graded each cuff as being intact, having a partial-thickness tear, or having a full-thickness tear. After a 3-week interval, each of the two observers then interpreted the T2*-weighted oblique coronal MR images in the same manner. Sensitivity and specificity for the two sets of interpretations were calculated in terms of the ability of the observers to use the two sequences to distinguish an intact cuff from a rotator cuff tear and to distinguish a partial-thickness cuff tear from an intact cuff or full-thickness tear. The results were then compared using Student's t test calculations.

For distinguishing an intact cuff from a torn cuff, the sensitivity of the observers' interpretations was lower for the T2*-weighted images than for the conventional dual-echo T2-weighted images. The specificity also decreased with interpretation of the T2*-weighted images for one observer and was unchanged for the second observer. For distinguishing a partial-thickness cuff tear from a non-partial-thickness cuff tear (intact or full-thickness tear), the sensitivity and specificity of both observers decreased with interpretations of the T2*-weighted images when compared with the conventional dual-echo T2-weighted images. The results did not reach statistical significance.

In this small study evaluating the ability of observers to diagnose rotator cuff tears on oblique coronal MR images only, interpretations of T2*-weighted images tended to be less sensitive and specific than interpretations of standard dual-echo T2-weighted images.