Original Research
The purpose of this study was to compare the diagnos-
tic performance of fat-saturation fast-spin-echo [FSE)
TO-weighted (T2W) sequences with conventional
spin-echo (CSE)T2W sequences in the detection of ro-
tator cuff pathology using surgery as the reference
standard. Oblique coronal dual-echo CSE and FSE T2W
images with fat saturation from 50 surgically con-
firmed MR shoulder examinations were acquired on a
1.5-T M R scanner. Blinded MR readers retrospectively
analyzed each imaging sequence separately and ulti-
mately correlated both sequences together with 6nd-
ings at surgery. FSE was 100% sensitive and 94%
specific in detection of full-thickness tears (n = 19)
and 73% sensitive and 97% specific in the detection
of partial-thickness rotator cuff tears [n = 13).There
was no statistically significant difference in the per-
formance of FSE with fat saturation compared with
CSE. The two discrepancies between imaging se-
quences related to the extent of partial-thickness
tears. Our findings suggest that fat-saturation FSE im-
aging can effectively replace CSE imaging in the eval-
uation of rotator cuff pathology.
Index terms: Fast spin-echo * MRI * Shoulder
JMRl 1997: 7:674-677
Abbreviations: CSE = conventional spin echo, FSE = fast spin echo, PDW
= pmton density weighted. STIR = short tau inversion recovery. TlW = T1
weighted, T2W = T 2 weighted.
From the Deparlmrnr of radio lo^, Memorial Kcgional Hospital, 3501 Johm
son Street, Hollywood. FL 33332. E-mail: MBZlalOaol.com. ReceivedJanuary
8. 1997: accepted March 10. Address correspondence to M.B.Z.
a ISMRM, 1997
CONVENTIONAL. SPIN-ECHO (CSE) imaging has been ac-
cepted widely as a n accurate method of evaluating rotator
cuff pathology (1-7). Fast spin echo (FSE)is a n attractive
alternative to CSE because the images produced provide
similar contrast to the familiar CSE technique in a frac-
tion of the time (8-1 1). A drawback or FSE shoulder im-
aging is that fat remains bright on T2-weighted (T2W)
FSE images, complicating identification of more subtle
rotator cuff defects and making it difficult to identify
small amounts of bursal fluid. The application of fat sup-
pression has been shown to enhance the conspicuity of
rotator cuff defects, subtle bursal inflammation, and
bone marrow edema while reducing chemical shift and
respiratory motion artifacts (4-6,12).
This study was performed to evaluate the diagnostic
performance of fat-saturation FSE T2W imaging in the
detection of surgically confirmed rotator cuff pathology
and to compare its efficacy with that of CSE.
MATERIALS AND METHODS
Our institution has supplemented fat-saturation FSE
images in the oblique coronal plane since the technique
became available at the end of 1992. We solicited clinical
records from 531 consecutive MR shoulder examinations
performed during this period and we were able to obtain
operative reports of 7 1 patients. Not all patients who were
imaged had symptoms that were referable to the cuff. Our
study group also contained patients with other shoulder
complaints, including instability. This potentially in-
creased the number of intact tendons in the study group
and also helped limit observer bias Tor tendon pathology.
Only patients who had surgical (open surgery or arthro-
scopic) confirmation of the status of the rotator cuff were
included in the study. In addition, patients with a history
of shoulder surgery (postoperative shoulders) were ex-
cluded. Fifty surgically confirmed shoulder MRIs made
up the final study population, which consisted of 18
women and 32 men, 28 to 84 years of age (mean, 51
years). The time between MRI and date of surgery ranged
from 1 to 198 days, with a mean of 55 days.
Oblique coronal images were acquired at 1.5-Tfield
strength (Signa, GE Medical Systems, Milwaukee, WI) with
a dedicated phased-may surface coil (MRI Devices, Wau-
kesha, WI]. Dual-echo spin-echo images were obtained
a.
b.
p’igure 1. Example of imaging concordance. Full-thickness ro-
tator cuff tear. (a] Oblique coronal CSE demonstrates a full-
thickness rotator ruff tear with tendon retraction (arrows]. Ib)
FSE with fat saturation depicts the same tear (arrows) with de-
creased scan time.
with TR/TE = 2300-2500/13-18,70 msec, 14-cm field of
view, 4-mm thickness with a 0.5-1.0-mm gap, two signal
averages, and a 256 X 128 matrix. The average imaging
time for the CSE series was 11 minutes and 15 seconds.
FSE images (effective TR/TE = 2500-3650/63-68 msec)
were obtained with €at saturation (Chem Sat, GE Medical
Systems), an echo train length of 8, 14-cm field of view, 4-
mm thickness with a 0.5-1.0-mm gap, two signal averages,
and a 256 X 192 matrix. The average imaging time for the
FSE sequence was 3 minutes. A higher matrix size was cho-
sen for the FSE sequence because greater resolution was
possible without a sigmficant increase in scan time.
a.
b.
Figure 2. Example of imaging concordance. Partial-thickness
tear. (a) CSE in a patient with a confirmed deep partial-thickness
tear of the undersurface of the supraspinatus tendon (arrow).@)
Note how conspicuous fluid in the tendon defect appears on this
FSE image with fat saturation (arrow).
Dual-echo CSE and single-echo T2W fat-saturation FSE
oblique coronal sequences were reviewed jointly by two ra-
diologists (S.D.N., M.B.Z.) who were experienced in mus-
culoskeletal imagmg and blinded to patient informationand
operative findings. Surgical reports were not reviewed until
after all of the MR images had been evaluated. Any discrep-
ant cases were re-reviewed in an attempt to idenbfy the
source of discordant findings. Results were subjected to
statistical analysis using x2 test.
For statistical purposes, the shoulders were subdivided
into three groups: no tear, partial tear, and complete tear.
Tendinosis and tendon surface fraying were placed into
the “no tear” category for statistical analysis. Partial tears
were diagnosed on MR when there was a defect in the
Volume 7 Number 4 JMRl 675
 Table 1
M R Diagnosis of Rotator Cuff Tears Versus Surgical
Findings
Sensitivitv Specificitv PPV N W
Complete tears at FSE 100% 94% 90% 100%
surgery(n= 19) CSE 100% 90% 86% 100%
Partial tears at. FSE 73% 97% 89% 93%
surgery In = 13) CSE 80% 97% 89% 95%
Note.-FSE, fast spin echo: CSE, conventional spin echo; PPV,
positive predictive value: NPV, negative predictive value.

tendon with focal fluid signal on T2W that extended to no

more than one tendon surface. Complete tears were di-
agnosed on MR when there was a defect in the tendon
that extended to both tendon surfaces and had fluid in-
tensity signal on T2W images.
lmages were obtained at the same section locations and
windowed in a similar fashion to allow valid comparison
of the two sequences. Image quality was subjectively as-
sessed as good, fair, or nondiagnostic.

RESULTS
FSE images revealed 21 complete tears, 10 partial a.
tears, and 19 intact tendons. CSE images revealed 22
complete tears, 10 partial tears, and 18 intact tendons.
At surgery, 19 patients had a complete tear, 13 patients
had a partial tear, and 18 patients had intact tendons.
There was no statistically significant difference (P> .8)
in the diagnostic performance between the FSE and CSE
sequences. With regard to surgically proven complete
tears [Fig. l),FSE was 100% sensitive and 94% specific,
compared with CSE which was 100% sensitive and 90%
specific. With regard to partial tears (Fig. 2), FSE was
73% sensitive and 97% specific, compared with CSE,
which was 80% sensitive and 97% specific. When com-
plete and partial tears were grouped together, FSE im-
aging had an overall sensitivity of 90% and a specificity
of 96Yo compared with surgical findings. CSE had a sen-
sitivity of 93% and a specificity of 94%. Table 1 summa-
rizes our statistical findings.
Two discrepancies between imaging sequences were
identified. One discrepancy occurred in a tendon with
significant morphologic alterations that, in addition, had
a subtle articular-sided partial-thickness tear diagnosed
on CSE and confirmed at surgery. This partial tear was
not identified on FSE images. The FSE images in this
case were judged to be only fair in technical quality, re- b.
lated to inhomogenous fat saturation and patient motion,
which may have contributed to the discordant finding. Figure 3. False-positive complete tear by CSE. (a) CSE sug-
The other imaging discrepancy also occurred in a tendon gested subtle areas of full-Chickness extension of a deep par-
with diffuse morphologic alteration. The CSE sequence tial-thickness tendon defect (arrows)within a diffusely abnormal
was interpreted as showing a subtle focus of full-thick- rotator cuff. Surgical exploration did not confirm the complete
tear. (a) The corresponding FSE sequence reveals the partial-
ness tearing that was seen neither on the FSE images nor thickness defects [arrows) and the diffusely abnormal cuff, but
at surgery (Fig. 3). Note also was made of two cases in the bursal surface appears intact, similar to what was found at
which tendon surface fraying described at surgery was surgery. Also note the marrow edema in the humeral head,
better appreciated on the dual-echo CSE than on the which is more prominent on this FSE imag?.
T2W FSE sequence.
With respect to the overall correlation with surgical volved the description of the depth of partial-thickness
findings, there was one case that was false-positive on tears. In two cases, intrasubstance partial-thickness
both CSE and FSE. This was diagnosed as a partial- tears were described on MRI but were not found at sur-
thickness tear at the articular surface, but the tendon gery. Because this is recognized to be a blind area for the
was considered normal at surgery. When this case was surgeon. these two latter cases were not labeled false-
reviewed, the MR images still were believed to reflect a positives.
partial tear. Other discrepancies with surgical findings Of the FSE images, 30 were judged to be of good quality
included two cases in which superficial partial-thickness and 20 were fair. Of the CSE images, 29 were judged to
tears were described as tendon surface fraying/ be of good quality and 2 1 were fair. No cases were deemed
irregularity at imaging. Two other discordant cases in- to be nondiagnostic in quality.

676 JMRl July/August 1997

 Subjectively, we felt that the improved spatial and con-
trast resolution of the FSE images with fat saturation of-
ten facilitated the recognition of tendon defects. The
higher resolution of the FSE images offered improved de-
piction of tendon defects, but the lower resolution CSE
images were equally capable of detecting tendon pathol-
ogy. The addition of fat saturation to the FSE images
made fluid appear more conspicuous, which assisted in
differentiating tears from tendinosis. Inhomogeneity of fat
suppression was apparent on many of the FSE images,
but there were no cases in which we believed that diag-
nostic quality was sacrificed.
DISCUSSION
FSE provides similar tissue contrast to CSE with
shorter imaging times. Recently published investigations
(13,14)have helped confirm the accuracy of FSE shoul-
der imaging compared with CSE. We found that the ad-
dition of fat saturation increases the conspicuity of fluid
within tendon defects and reduces chemical shift and
respiratory motion artifact. Our data show no statistically
significant difference in detection and characterization of
rotator cuff defects between T2W FSE images with fat sat-
uration and T2W CSE imaging. Diagnostic performance
of both sequences was consistent with data presented
previously in the literature [l-7). Both sequences per-
formed better in the diagnosis of complete tears than par-
tial tears. Subjectively, the improved image resolution
and increased contrast of the FSE sequence facilitated
the recognition of tears.
T2W sequences are the most commonly used sequence
in the detection of rotator cuff tears. Proton-densitj-
weighted (PDW) and T1-weighted (TlWJ sequences also
are useful because they allow better visualization often-
dinosis and subtle morphologic alterations of the tendon
surface, although the clinical implications of these find-
ings are still in debate. Because we used only a single
late-echo FSE sequence, we were less able to detect these
latter changes and thus did not include them in our sta-
tistical analysis. If the CSE T2W sequence is abandoned,
we suggest the use of a short TE sequence (either CSE
T l W or FSE PDW to continue detecting these tendon
changes.
Implementation of this FSE pulse sequence is easiest
with MRI scanners that can reliably generate fat satura-
tion. Inhomogeneous fat saturation or inadvertent satu-
ration of the water peak may potentially yield nondi-
agnostic images. A similar investigation employing FSE
short tau inversion recovery (STIR)to evaluate the rotator
cuff would certainly be of value, especially for low and
midfield strength MR scanners.
In conclusion, the FSE T2W fat-saturation sequence
seems to perform well in the detection of rotator cuff de-
fects. Compared with CSE, FSE allows the use of larger
matrices and may decrease patient motion and respira-
tory artifacts. The use of Faster imaging sequences has
the additional benefit of potentially increasing patient
throughput. In conjunction with fat saturation, our find-
ings suggest that T2W FSE imaging may be able to re-
place CSE imaging in the evaluation of rotator cuff tearsl.
References
1. Farley TE, Neumann CH, Steinbach LS,Jahnke AJ, Petersen
SS. Full-thickness tears of the shoulder: diagnosis with MR
imaging. h l R Am J Roentgenol 1992: 158:347-351.
2. Iannotti JP, Zlatkin MB, Esterhai J L , et al. Magneticresonance
imaging ofthe shoulder. J Bone Joint Surg Am 1991; 73: 17-29.
3. Kjellin J , Ho CP, Cenfla, et al. Alteration of the supraspinatus
t.endon a t MR imaging: correlation with histopathologic findings
in cadavers. Radiology 1991; 181:837-841.
4. Mirowitz SA. Normal rotator cuff: MR imaging with conven-
tional and fat-suppression techniques. Radiology 1991: 180:
735-740.
5. Quinn SF, Sheley RC, Demlow TA, Szumowski J . Rotator cuff
tendon tears: evaluation with fat-suppressed MR imaging with
arthroscopic correlation in 100 patients. Radiology 1995; 195:
497-50 1.
6. Traughber PD, Goodwin TE. Shoulder MRI: arthroscopic cor-
relation with emphasis on partial tears. J Comput Assist Tomogr
1992; 16:129-133.
7. Zlatkin MB, Iannotti JP, Roberts MC, et a]. Rotator cuff tears
diagnostic performance of MR imaging. Radiology 1989: 172.
223-229.
8. Constable RT, Anderson AW, Zliong J, Gore JC. Factors influ-
encing contrast in fast spin-echo MR imaging. Magn Reson Im-
aging 1992; 10:497-511.
9. Rubin DA, Kneeland JB, Listerud J, UnderbergDavis ST.
Dalinka MK. MR diagnosis of meniscal tears of the knee: value
of iast spin-echo vs. conventional spin-echo pulse sequences,.
AJRAm J Roentgenol 1994; 1621131-1135.
10. Jones KM, Mulkern RV, Schwartz RB, Oshio K, Barnes PD,
Joelsz FA. Fast spin-echo MK imaging of the brain and spine:
current concepts. AJRAm J Roentgenol 1992; 158:131%1320.
11. Smith RC, Rcinhold C, Lange RC, McCauley TR. Kier R, McCarthy
S. Fast spin-echo MR imaging of the female pelvis: Part I. Use
of a whole-volume coil. Radiology 1992: 184:665-669.
12. Reinus WR, Shady KL, Mirowitz SA, Totty WG. MR diagnosis
of rotator cuff tears of the shoulder: value of usingT2-weighted
fat-saturated images. AJR Am J Roentgenol 1995: 164:1451-
1455.
13. Singson Ru, Hoang T, Dan S, Friedman M. MR evaluation of
rotator cuff pathology using T2-weighted fast spin-echo tech-
nique with and without fat suppression. AJR Am J Roentgenol
1996: 166:1061-1065.
14. Sonin AH, Pcduto AJ, Fitzgerald SW, Callahan CM, Breslet
ME. MR imaging of the rotator cuff mechanism: comparison of
spin-echo and turbo spin-echo sequences. AJR Am J Roent-
genol 1996; 167:333-338.
Volume 7 Number 4 JMRl 677