Practical assessment of rotator cuff muscle volumes using shoulder MRI

Abstract

Reliable quantitative assessment of rotator cuff (RC) muscle volumes can be done by reconstructing multiple MRI images of the entire shoulder. However, an equally reliable, but less time-consuming, method is needed for clinical practice. We compared the only method reported for estimation of volume with a new simple MRI technique. Both methods were validated by multiple MRI image reconstruction. We performed MRI scans of 10 cadaver shoulders and determined the cross-sectional areas of RC muscles with two methods, using image analyzing software. In Method 1, the cross-sections were determined on a single image, previously described as a Y-shaped image. In Method 2, the cross-sections were calculated from two images: the Y-shaped and an image located medially, twice the distance from the glenoid articular surface compared to the Y-shaped image. We compared the results of these two methods with the volume of multiple MRI image reconstruction, which took into account all images consisting of RC cross-sections. Pearson correlations for Method 1 were 0.96, 0.94 and 0.75, and for Method 2, 0.96, 0.97 and 0.93 for the supraspinatus, infraspinatus/teres minor and subscapularis muscle volumes when compared with the volumes determined by the multiple image reconstruction method. The Bland-Altman method showed better agreement with multiple MRI image reconstruction, using Method 2, to determine supraspinatus, infraspinatus, and subscapularis muscle volumes (p<0.001 for each). The mean intra- and inter-observer variabilities of Method 1 was 3.9% and 2.9% and that of Method 2, 3.0% and 1.7%, respectively. Both methods can be used for quantitative assessment of RC muscle volumes. However, Method 2, using two easily reproducible MRI images is more accurate for the evaluation of the supraspinatus and infraspinatus/teres minor muscles and particularly for the subscapularis muscle.