Currently, the fatty degeneration of the rotator cuff muscles, which has a major influence on the outcome after rotator cuff repair [1], is often investigated by visual inspection of T$$$_1$$$-weighted MR images following the (modified) Goutallier classification [2]. The validity of this rather subjective technique is in debate and several studies showed very different results concerning the inter- and intraobserver reliability [3,4].

There is thus a need for more quantitative measurements to correctly assess fatty degeneration of the supraspinatus muscle after rotator cuff tear. Previous studies showed that it is possible to absolutely quantify the amount of fat inside an arbitrarily shaped ROI in the shoulder using the 2D spectroscopic fast low angle shot (SPLASH)-technique [5,6]. However, measurement-time of this spectroscopic technique is rather long compared to different MRI-techniques like, for example, T$$$_1$$$ mapping for tissue characterization.

Shear wave ultrasound, which is a cost-effective methodology, provides an indirect measure of the tissue’s elasticity via the shear wave propagation speed which might be influenced by the amount of fat in the observed region [7]. However, a direct measure of the fat fraction is not possible and results may be corrupted by the amount of overlaying soft tissue.

Thus, it was the goal of the present study to compare three quantitative techniques for investigation of the fatty degeneration of the supraspinatus muscle: 2D SPLASH-MRI, model-based acceleration of parameter mapping (MAP) [8] to determine T$$$_1$$$ and shear wave ultrasound.

22 patients after rotator cuff tear underwent T$$$_1$$$-weighted MRI, SPLASH-MRI and MAP at 3T (Magnetom Skyra, Siemens, Erlangen, Germany) and shear wave ultrasound of the supraspinatus muscle. For SPLASH, 21 spoiled gradient echo images with echo times from 5ms to 25ms were acquired (TR: 35ms, flipangle: 10°, FOV: 278x278mm$$$^2$$$, matrix: 128x128, slice thickness: 5mm, Taq: 126s). A ROI was placed in the supraspinatus muscle using MATLAB (R2014b, The MathWorks, Natwick, MA, United States). The signal fat fraction was then obtained by a temporal Fourier transform, followed by an AMARES fit [9] in jMRUI [10,11].

T$$$_1$$$ maps of the same image slices were acquired using the previously proposed MAP-algorithm [8]. This technique iteratively applies an exponential signal model to a radially acquired inversion-recovery prepared FLASH acquisition, generating T$$$_1$$$ maps in very short acquisition times (TR: 4.2ms, flip angle: 7°, FOV: 280x280mm$$$^2$$$, matrix: 128x128, slice thickness: 5mm, Taq: 4s). To obtain the area fat fraction from the calculated T$$$_1$$$ maps pixels in the supraspinatus muscle were classified either as fat or water depending on their T$$$_1$$$ value by means of a simple threshold.

Ultrasound was performed using a Siemens Acuson S3000 (Siemens, Erlangen, Germany) device. Tissue elasticity was measured by aligning the transducer parallel to the muscle fibers at the largest diameter of the supraspinatus muscle and calculating the median of the velocities measured at 10-15 points.

The obtained fat fractions from the two MRI-based techniques were compared using a Wilcoxon matched pairs test and Pearson’s correlation coefficient was calculated. To not restrict the effect to a linear relationship between MR parameters and shear wave propagation speed Spearman’s rank correlation test was used to compare the results.

Figure 1 shows an example of a T$$$_1$$$-weighted image, the corresponding SPLASH-spectrum as well as the T$$$_1$$$ map of the same slice.

Obtained fat fractions from MAP (mean: 19%±13%; median: 17%; min: 1%; max: 50%) did not differ significantly (p=0.22) from those obtained with SPLASH (mean: 17%±14%; median: 13%; min: 0%; max: 48%). Pearson’s correlation coefficient was r = 0.82. Figure 2a shows the fat fraction values obtained with SPLASH against those obtained with MAP.

Spearman’s correlation coefficient between shear wave ultrasound and the MRI-based techniques was 0.30 for SPLASH and 0.19 for T$$$_1$$$ mapping. Figure 2b presents the shear wave propagation speed against the fat fractions from SPLASH and T$$$_1$$$ mapping.

The results of the T$$$_1$$$ mapping technique are in good concordance with the results of the SPLASH quantification which served as gold standard in this study. In contrast, the obtained values from shear wave ultrasound do not correlate well with these two quantitative MRI-based techniques.

Since T$$$_1$$$ mapping using the MAP-algorithm only requires a 4 seconds acquisition, this technique seems very promising for future applications.