Skeletal Radiol (1995) 24:495-498 Skeletal Radiology Effect of field of view on MR diagnosis of rotator cuff tears Michael J. Tuite, M.DJ, Donald R. Yandow, M.DJ, Arthur A. De Smet, M.D. 1, John E Orwin, M.D. 2, Fernando A. Quintana, B.S. 3 1 Department of Radiology, University of Wisconsin Clinical Science Center, Madison, Wisconsin, USA 2 Division of Orthopedic Surgery, University of Wisconsin Clinical Science Center, Madison, Wisconsin, USA 3 Department of Biostatisties, University of Wisconsin Clinical Science Center, Madison, Wisconsin, USA Abstract. Purpose. To determine the effect of decreas- ing the field of view (FOV) on the accuracy of MR for diagnosing rotator cuff tears. Material and methods. One hundred shoulder MR scans with surgical correlation were evaluated for the presence or absence of a cuff tear. The sensitivity and specificity of MR relative to the surgical results were determined for the 59 patients scanned with a 24-cm FOV, and the 41 patients scanned with an 18-cm FOV. All other imag- ing parameters including acquisition time were identical. The sensitivity and specificity of the two groups were compared using a t-test. Results. The specificity of MR for diagnosing a rotator cuff tear improved from 0.65 for the 24-cm FOV group to 0.89 for the 18-cm FOV group (P=0.04). The sensitiv- ity changed from 0.91 to 0.96 (P=0.25). Conclusion. Reducing the FOV from 24 cm to 18 cm re- sults in a statistically significant improvement in speci- ficity of MR for diagnosing rotator cuff tears. Key words: Shoulder injuries - Tendon injuries - Mag- netic resonance Field of view (FOV) is one of the most effective opera- tor-controlled parameters determining spatial resolution in MR imaging [1-3]. Minor reductions in FOV cause significant improvements in spatial resolution, without prolonging the acquisition time [1]. One of the disadvan- tages of decreasing the FOV, however, is a reduction in the signal-to-noise (S/N) ratio [1]. S/N ratio and spatial resolution both contribute to image quality, one of the factors that determines the diagnostic accuracy of MR [1, 3, 4]. Recent papers on MR of the knee and temporoman- dibular joint (TMJ) have shown that increases in spatial Correspondence to: M. J. Tuite, Department of Diagnostic Radi- ology, University of Wisconsin Clinical Science Center, 600 Highland Avenue, Madison, WI 53792-3252, USA resolution can lead to an overall improvement in diag- nostic accuracy and anatomic detail [5, 6]. Attempts to improve spatial resolution in shoulder MR, however, have usually caused authors to increase the number of signals averaged (NEX) to maintain an acceptable S/N ratio [7, 8]. This further prolongs what are already often lengthy shoulder MR examination times [9]. The data for this study were collected as part of a separate project evaluating our ability to grade the size of rotator cuff tears. When reviewing some of the images after correlating our MR grades with the arthroscopic findings, we noted that most of our false-positives were obtained with a 24-cm FOV. The purpose of this study was to determine the effect of a decrease in the FOV on the sensitivity and specificity of MR for diagnosing rota- tor cuff tears. Subjects and methods We identified 100 consecutive patients at our institution who had been given a shoulder MR examination and arthroscopy by a sin- gle orthopedic surgeon. The orthopedist is subspecialty trained in shoulder surgery and has a practice specializing in shoulder dis ease. All scans were performed on a 1.5-'1"scanner (GE, Milwaukee, Wisc). with a 7 in. single-loop dedicated shoulder coil (GE). Pa- tients were positioned supine with the humerus in comfortable, neutral rotation. Pulse sequences for each patient were: oblique coronal Tl-weighted 600/20 (TR/TE), 1 NEX; oblique coronal gradient echo 550/20 (TRfFE), 30 ~ flip angle, 4 NEX; and oblique sagittal T2-weighted 2000/20, 90 (TR/TE), 1 NEX. The oblique coronal and sagittal images were localized perpendicular and par- allel, respectively, to the glenoid. Matrix size was 256• (fre- quency • phase encoding) for all sequences except the oblique sagittal, which was 256• Slice thickness was 4 mm for all se- quences. The FOV was 24 cm for 59 of the patients and 18 cm for the other 41 patients. The scans obtained at a 24-cm FOV were per- formed between 11 October 1990 and 8 November 1991, and those with an 18-cm FOV were performed between 9 November 1991 and 4 February 1993. The average age was 43 years in the 24-cm FOV group, and 41 years in the 18-cm FOV group. There were 19 females and 40 males in the first group, and 10 females and 31 males in the second. The images from the patients scanned 9 1995 International Skeletal Society