Musculoskeletal Imaging Original Research

Oh et al.
Conventional Versus Isotropic MR Arthrography of the Shoulder

Musculoskeletal Imaging
Original Research

Comparison of Indirect Isotropic

MR Arthrography and Conventional
Downloaded from www.ajronline.org by 202.159.8.37 on 02/18/17 from IP address 202.159.8.37. Copyright ARRS. For personal use only; all rights reserved

MR Arthrography of Labral Lesions

and Rotator Cuff Tears:
A Prospective Study
Dae Kun Oh1 OBJECTIVE. The purpose of our study was to prospectively compare the diagnostic ac-
Young Cheol Yoon1 curacy of 3D isotropic indirect MR arthrography with conventional sequences of indirect MR
Jong Won Kwon1 arthrography for the diagnosis of labral and rotator cuff lesions on a 3-T MR unit.
Sang-Hee Choi1 SUBJECTS AND METHODS. Thirty-six consecutive patients who were scheduled for
Jee Young Jung1 shoulder arthroscopic surgery at our institution underwent indirect MR arthrography. Both
conventional sequences and an additional 3D isotropic sequence were obtained 1 day before
Sooho Bae1
arthroscopic surgery. Two musculoskeletal radiologists prospectively evaluated the images in
Jaechul Yoo 2 consensus for the presence of superior and anterior labral lesions and subscapularis and su-
Oh DK, Yoon YC, Kwon JW, et al. praspinatusinfraspinatus tendon tears using the conventional sequences and the 3D isotropic
sequence. We analyzed the statistical difference between the sensitivities and specificities of
both methods using arthroscopic findings as the reference standard.
RESULTS. Surgical findings confirmed the presence of 23 superior labral lesions, eight
anterior labral lesions, 21 subscapularis tears, and 24 supraspinatusinfraspinatus tears. The
sensitivity and specificity of the conventional sequences were 74% and 54% for superior
labral lesions, 88% and 96% for anterior labral lesions, 67% and 85% for subscapularis ten-
don tears, and 96% and 75% for supraspinatusinfraspinatus tendon tears. The sensitivity and
specificity of the 3D isotropic sequence were 70% and 85% for superior labral lesions, 100%
and 100% for anterior labral lesions, 67% and 85% for subscapularis tendon tears, and 96%
and 67% for supraspinatusinfraspinatus tendon tears. No statistically significant difference
was seen in sensitivities and specificities for both methods.
CONCLUSION. Three-dimensional isotropic MR arthrography sequences with multi-
planar reconstruction can provide a similar capability for the diagnosis of labral and rotator
cuff lesions as conventional MR arthrography sequences but in a shorter imaging time.

Keywords: arthrography, MR; MRI, 3D; MRI, high field
strength; MRI, rapid imaging; shoulder
DOI:10.2214/AJR.08.1223
Received May 12, 2008; accepted after revision
July 30, 2008.
1
Department of Radiology, Samsung Medical Center,
School of Medicine, Sungkyunkwan University, 50
Ilwon-dong, Kangnam-ku, Seoul, 135-710, Republic of
Korea. Address correspondence to Y. C. Yoon (ycyoon@
skku.edu).
2
Department of Orthopedic Surgery, School of Medicine,
Sungkyunkwan University, Seoul, Republic of Korea.
AJR 2009; 192:473479
0361803X/09/1922473
American Roentgen Ray Society
I
n the diagnosis of disorders of the
glenoid labrum and rotator cuffs,
indirect and direct MR arthrog-
raphy has been reported to be su-
perior to conventional MRI, particularly for
superior labral anterior to posterior (SLAP)
lesions and for partial-thickness rotator cuff
tears [15]. In addition, the recently intro-
duced 3-T MRI systems have provided an in-
creased signal-to-noise ratio (SNR) for use
when it is feasible to achieve faster and high
resolution. Faster imaging will lead to de-
creased motion artifacts, increased conve-
nience of the patient, and improved and more
accurate diagnosis because of increased reso-
lution [6]. Recently, isotropic imaging that al-
lowed saving imaging time by using a single
acquisition and by obtaining diverse images
using reformation in arbitrary planes has been
tested as an alternative method [7]. In retro-
spective studies, Magee and Williams [8, 9]
reported that shoulder imaging on a 3-T sys-
tem showed high accuracy for the diagnosis of
supraspinatus tendon and labral lesions; and
isotropic imaging using a fast gradient on a
3-T system showed labral lesions and rotator
cuff lesions as well as conventional imaging
in a short time [10]. To our knowledge, no pro-
spective study has described the usefulness of
isotropic imaging of the shoulder. The pur-
pose of this study was to compare prospec-
tively the diagnostic accuracy of 3D isotropic
MR arthrography using a gradient-refocused
echo (GRE) technique with conventional
MR arthrography for the diagnosis of labral
and rotator cuff lesions on a 3-T MR unit.

AJR:192, February 2009 473


Subjects and Methods
Patients
The institutional review board approved the
study, which was conducted in compliance with
HIPAA regulations. Patient informed consent
was obtained.
Between March 2006 and June 2006, 36 con
Downloaded from www.ajronline.org by 202.159.8.37 on 02/18/17 from IP address 202.159.8.37. Copyright ARRS. For personal use only; all rights reserved
secutive patients who were scheduled for shoulder
arthroscopic surgery at our institution were en
rolled in this prospective study. Arthroscopic
surgery was indicated on the basis of clinical
symptoms and signs and MRI findings. The study
population consisted of 16 men (age range, 2071
years; mean age, 47.1 years) and 20 women (age
range, 2777 years; mean age, 59.3 years). Six
patients presented with shoulder instability and
the remaining 30 patients with shoulder pain and
motion limitation. Exclusion criteria were a his
tory of shoulder surgery, surgery for infection or
tumor, and contraindications to MRI.
MRI Protocol
All MR examinations were performed 1 day
before arthroscopic surgery. According to our
standard protocol, 0.1 mmol/kg of gadopentetate
dimeglumine (Magnevist, Bayer Schering Pharma)
was injected IV, and patients were instructed to
exercise for 15 minutes. After active exercise,
MRI was performed with a 3-T whole-body MR
scanner (Gyroscan Intera Achieva, Philips Health
care) with a dedicated receive-only shoulder coil.
Patients were positioned with the humerus in a
neutral position and the thumb pointing upward.
The following conventional MR arthrography
sequences were obtained: Fat-suppressed T1-
weighted fast spin-echo sequences were obtained
in the axial plane (TR range/TE range, 434
565/1824; section thickness, 3 mm; field of view,
15 cm; matrix dimensions, 224 224); in the
coronal oblique plane parallel to the long axis of
the supraspinatus tendon (434565/2024; section
thickness, 3 mm; field of view, 15 cm; matrix,
224 224); and in the sagittal oblique plane
perpendicular to the long axis of the supraspinatus
tendon (434561/1824; section thickness, 4 mm;
field of view, 15 cm; matrix dimensions, 256
256). T2-weighted fast spin-echo sequences were
obtained in the axial plane (TR range/TE, 2,868
3,184/80; section thickness, 3 mm; field of view, 15
cm; matrix dimensions, 224 224) and in the
coronal oblique plane (2,6612,906/80; section
thickness, 3 mm; field of view, 15 cm; matrix
dimensions, 224 224). The number of excitations
for conventional MR arthrography was 2. The
echo-train length of the T2-weighted fast spin-echo
sequence was 16. The total scanning time for
conventional MR arthrography was 16 minutes 40
seconds. Subsequently, a 3D fast GRE technique
474 AJR:192, February 2009
Oh et al.
with fat suppression (THRIVE: T1-weighted high-
resolution isotropic volume examination) was
performed in the axial plane to obtain additional
3D isotropic MR images using the following
imaging parameters: TR/TE, 7.8/3.4; section
thickness, 0.6 mm; field of view, 18 cm; matrix
dimensions, 300 300; voxel size, 0.6 0.6 0.6
mm; sensitivity encoding (SENSE), 2; number of
excitations, 2; flip angle, 7; number of slices, 120.
The total scanning time was 5 minutes 32 seconds.
Image Analysis
MR images were prospectively evaluated by
two musculoskeletal radiologists, one with 5 years
of experience in musculoskeletal MRI and one
with 2 years of experience in musculoskeletal
MRI, by consensus. We analyzed labral lesions,
subscapularis tendon tears, and supraspinatus
infraspinatus tendon tears. The criteria for de
fining a labral lesion were as follows [11, 12]:
identification of contrast material extending into a
linear or complex tear cleft in the labrum; the
absence of the labrum; a marked deformity of the
labrum; intralabral high signal intensity reaching
the articular surface of the labrum; truncation or
fragmentation of the labrum; and displacement of
the labrum from its expected anatomic location.
These labral lesions were evaluated and were re
corded as present or absent according to location.
In addition, the following imaging features were
used to differentiate between a SLAP type II lesion
and a sublabral recess. Lateral or superior extension
of contrast medium into the superior labrum and
the biceps anchor indicated a SLAP type II lesion,
whereas medial extension of the contrast medium
with a smooth linear appearance between the
superior labrum and the glenoid rim was indicative
of a sublabral recess [13]. A Buford complex con
sisting of an absent anterosuperior labrum and a
thick cordlike middle glenohumeral ligament [14]
that may be mistaken for a displaced labral fragment
on arthrography was excluded [15]. Subscapularis
tendon tears were defined as follows: discontinuity
of the tendon, contrast medium entering the ten
don, abnormal signal intensity, and caliber change.
Furthermore, ancillary signs of subscapularis
tendon abnormality were considered [16]. These
signs included the presence of fatty infiltration in
the subscapularis muscle and abnormalities in the
course of the long biceps tendon. A supraspinatus
infraspinatus tear was graded as either a full-thickness
tear or a partial-thickness tear. A partial-thickness
tear was classified as an articular surface or bursal
surface tear. The imaging criterion of a full-
thickness tear was complete discontinuity in the
tendon. The criteria for partial-thickness articular
surface tears included focal discontinuity of the
undersurface of the supraspinatusinfraspinatus
tendon. The criteria of partial-thickness bursal
surface tears were abnormal signal intensity of the
bursal surface of the tendon or focal disr uption of
the bursal surface of the tendon [17].
Two observers evaluated labral lesions, sub
scapularis tendon tears, and supraspinatusinfra
spinatus tendon tears with conventional MR
arthrography sequences (method A). The evalu
ation was repeated after 2 weeks with the 3D
isotropic MR arthrography sequence (method B).
Images were presented in random order at each
reading session and were evaluated using a PACS
(Centricity Radiology RA 1000, GE Healthcare).
The reformation in method B was performed
simultaneously during image analysis using com
mercially available software (Advantage Windows
suite 1.0, GE Healthcare) according to oblique
coronal and oblique sagittal plane images and
additional images in arbitrary planes if needed.
Arthroscopic Surgery
An arthroscopic finding was considered the
reference standard. One orthopedic surgeon with
6 years of experience in shoulder surgery who was
not blinded to the indirect MR arthrographic
images performed the arthroscopic surgery. The
surgeon recorded the presence or absence of a
superior labral lesion, including significant de
generation and fraying requiring treatment. Anter
ior labral lesions, subscapularis tendon tears, and
supraspinatusinfraspinatus tendon tears were re
corded in the same manner.
Statistical Analysis
The sensitivity, specificity, and accuracy of both
method A and method B were calculated. For
calculation of the sensitivity, specificity, and
accuracy of supraspinatusinfraspinatus tendon
tears, grades of the tear (a full- or partial-thickness
tear) were not considered. Sensitivity for each type
of supraspinatusinfraspinatus tendon tear as
determined by both methods was also calculated.
We analyzed the statistical difference of the sen
sitivities, specificities, and accuracies in diagnos
ing labral lesions, subscapularis tendon tears, and
supraspinatusinfraspinatus tendon tears, as well
as the sensitivities for each type of supraspinatus
infraspinatus tendon tear for methods A and B using
the McNemar test. A p value of less than 0.05 was
considered a statistically significant difference.
To calculate the adequate sample size, we
hypothesized that sensitivity and specificity for
the diagnosis of labral and rotator cuff lesions
using method B would be equal to those of method
A on a 3-T MR unit. If a difference in sensitivity
and specificity of method A versus method B was
less than 20%, it was regarded as not meaningful.
We assumed that the sensitivity of method A for the
 Downloaded from www.ajronline.org by 202.159.8.37 on 02/18/17 from IP address 202.159.8.37. Copyright ARRS. For personal use only; all rights reserved

TABLE 1: Findings with Conventional MR Arthrography (Method A), 3D Isotropic MR Arthrography (Method B), and Arthroscopic Surgery
Superior Labral Lesion Anterior Labral Lesion Subscapularis Tendon SSPISP Tendon
Patient Age
No. Sex (y) Method A Method B Surgery Method A Method B Surgery Method A Method B Surgery Method A Method B Surgery
1 M 62 Tear Tear Tear Tear Tear PT articular
2 F 76 Tear Tear Tear Tear FT FT FT
3 F 69 Tear Tear Tear Tear Tear FT FT FT

AJR:192, February 2009

4 F 69 Tear Tear Tear Tear Tear FT FT FT
5 F 72 Tear Tear Tear Tear FT FT FT
6 F 53 Tear Tear Tear PT articular
7 M 57 Tear Tear PT bursal PT bursal PT bursal
8 F 27 Tear Tear Tear Tear Tear
9 M 57 Tear Tear Tear Tear Tear FT PT bursal PT articular, PT bursal
10 F 57 Tear Tear PT articular PT articular
11 F 64 Tear Tear PT articular PT bursal PT articular, PT bursal
12 M 71 Tear Tear Tear Tear Tear Tear Tear Tear FT FT FT
13 F 62 Tear Tear Tear Tear Tear Tear Tear Tear Tear FT FT FT
14 F 59 Tear Tear FT FT FT
15 F 67 Tear Tear Tear PT articular PT articular PT articular
16 M 54 Tear Tear Tear PT articular PT articular PT articular
17 M 50 Tear Tear Tear Tear PT articular PT articular
18 M 60 Tear Tear Tear Tear Tear Tear FT FT FT
19 M 56 Tear Tear Tear Tear Tear PT bursal PT bursal PT bursal
20 F 77 Tear Tear Tear Tear Tear FT FT FT
21 F 49 Tear Tear Tear Tear Tear Tear PT bursal PT bursal PT bursal
22 F 67 Tear Tear Tear Tear Tear FT FT FT
23 F 43 Tear Tear Tear Tear PT bursal
24 F 77 Tear Tear Tear Tear Tear Tear FT FT FT
25 M 53 Tear PT bursal PT bursal FT
26 M 22 Tear Tear Tear Tear Tear Tear
27 M 25 Tear Tear Tear Tear Tear
Conventional Versus Isotropic MR Arthrography of the Shoulder

28 F 46 Tear Tear
29 M 20 Tear Tear Tear Tear
30 F 50 Tear PT bursal
31 M 41 Tear Tear Tear
32 M 57 Tear Tear FT FT FT
33 F 59 Tear Tear Tear FT FT FT
34 F 43 Tear Tear Tear Tear PT articular
35 M 47 PT bursal FT PT bursal
36 M 22 Tear Tear PT articular PT articular PT articular

475
NoteSSPISP = supraspinatusinfraspinatus, FT = full-thickness, PT = partial-thickness.
 Oh et al.

diagnosis of rotator cuff tear was 80%. An error

level or confidence level of 5% and a error level
or statistical power (1 ) of 80% were used. The
calculated sample size was 36 (SPSS, version 12).

Results
Patient data and results are summarized in
Downloaded from www.ajronline.org by 202.159.8.37 on 02/18/17 from IP address 202.159.8.37. Copyright ARRS. For personal use only; all rights reserved

Table 1. After arthroscopic surgery, 23 pa-

tients were identified as having superior
labral lesions, eight as having anterior labral
lesions, 21 as having subscapularis tendon
tears, and 24 as having supraspinatusin-
fraspinatus tendon tears. Among supraspina-
tusinfraspinatus tendon tears, there were 14
full-thickness tears in 14 patients, six articu- A B
lar surface partial-thickness tears in six pa-
Fig. 1Arthroscopically proven superior labral lesion in 53-year-old woman.
tients, and six bursa surface partial-thickness A, Fat-suppressed T1-weighted oblique coronal image shows increased signal intensity between superior
tears in six patients. Two patients had both labrum and glenoid rim (arrow), which was interpreted as superior labral lesion.
articular and bursal surface partial-thickness B, Three-dimensional isotropic MR arthrography sequence with oblique coronal reformatted image shows
similar findings (arrow). Lesion was interpreted as superior labral lesion.
tears simultaneously.
Seventeen of 23 patients with superior
labral lesions were correctly diagnosed using
method A and 16 patients using method B
(Fig. 1). Superior labral lesions in two pa-
tients were missed using method A only, and
lesions in another three patients were missed
using method B only. Superior labral lesions
in four patients were missed using both
methods A and B, and a patient with no supe-
rior labral lesion detected at arthroscopy was
misdiagnosed as having a superior labral le-
sion using both methods. Seven of eight pa-
tients with anterior labral lesions were cor-
rectly diagnosed using method A, and all
eight patients were correctly diagnosed using
method B (Fig. 2). One false-positive diag- A B
nosis each and one false-negative diagnosis Fig. 2Arthroscopically proven anterior labral lesion in 22-year-old man with recurrent shoulder dislocation.
each were generated using only method A. A, Fat-suppressed T1-weighted axial image shows contrast material extending into bony labrum (arrow), which
was interpreted as anterior labral lesion.
Fourteen of 21 patients with subscapularis B, Axial 3D isotropic MR arthrography image again shows lesion (arrow), which was interpreted as anterior
tendon tears were correctly diagnosed using labral lesion.
methods A and B. Four false-negative diag-
noses were generated using both methods For lesion-to-lesion evaluation of supraspi- an articular surface tear was missed, and an-
(Fig. 3). Each method showed three false- natusinfraspinatus tendon tears, the detec- other tear was regarded as a full-thickness
positive diagnoses. Twenty-three of 24 pa- tion rate of both methods for a full-thickness tear by method A. Overall detection rates for
tients with a supraspinatusinfraspinatus tear was 100% (14/14). A full-thickness tear, supraspinatusinfraspinatus tears on a le-
tendon tear were correctly diagnosed using however, was reported as a partial-thickness sion-to-lesion basis were 92% with both
methods A and B (Fig. 4). One false-negative tear by both methods. For partial-thickness methods. With method B, another partial-
diagnosis was generated by both methods. articular surface tears, the detection rates thickness bursal surface tear was considered
Method A showed three false-positives, and were 83% (5/6) for method A and 50% (3/6) a full-thickness tear. The overall detection
method B, four false-positives. Among these for method B (Fig. 4). Two partial-thickness rates for supraspinatusinfraspinatus tendon
false-positive diagnoses, two were misdiag- articular surface tears that coexisted with a tears were 92% (24/26) for method A and
nosed by both methods (Fig. 5). Diagnostic bursal surface tear were missed by method 88% (23/26) for method B. This difference
values of both methods for evaluating the B. In addition, one of the tears was thought to was not statistically significant.
labral abnormalities and rotator cuff tears be a full-thickness tear using method A. For
are summarized in Table 2. No statistically partial-thickness bursal surface tears, the de- Discussion
significant difference was seen between the tection rate was 83% (5/6) for method A and Three-dimensional Fourier transforma-
diagnostic efficacy of method A and method 100% (6/6) for method B. A partial-thick- tion imaging techniques using GRE sequenc-
B for a labral lesion and rotator cuff tear. ness bursal surface tear that coexisted with es have been applied to various structures,

476 AJR:192, February 2009

 Conventional Versus Isotropic MR Arthrography of the Shoulder

including cartilage of the knee and the liga-

ments of the wrist and ankle. The use of
these techniques has several advantages,
such as high spatial resolution and the oppor-
tunity of postprocessing [1820]. In addition,
a recent study showed that a combination of
higher field strength and a parallel imaging
Downloaded from www.ajronline.org by 202.159.8.37 on 02/18/17 from IP address 202.159.8.37. Copyright ARRS. For personal use only; all rights reserved

technique led to improved diagnostic ability

and a reduction of scanning time in imaging
of the ankle [21]. THRIVE, which was used
in this study, is a T1-weighted turbo field
echo 3D scan with spectral attenuated inver-
sion recovery (SPAIR) fat suppression using
sensitivity encoding (SENSE). The ability of
THRIVE for isotropic voxel imaging permits
A B multiplanar reconstruction (MPR) without
Fig. 3Arthroscopically proven subscapularis tendon tear in 57-year-old man. loss of image conspicuity. Its T1 nature and
A and B, Fat-suppressed T1-weighted (A) and 3D isotropic MR arthrography (B) axial images show subtle high its faster image acquisition because of its
signal intensity at articular side of cranial portion of subscapularis tendon (arrows), which was interpreted as
normal subscapularis tendon.
short TE (35 milliseconds) and TR (710
milliseconds) (12.5 seconds per slice) are
other features that made THRIVE suitable
for MR arthrography in this study.
In this study, although no statistically sig-
nificant difference was seen, the sensitivity
and specificity of the 3D isotropic MR se-
quence for the diagnosis of anterior labral le-
sions and the specificity for the superior
labrum were slightly higher than the sensitiv-
ity and specificity of conventional MR arth
rography sequences. The sensitivity for a su-
perior labral lesion on a 3D isotropic MR se-
quence was slightly less than that reported in
a previous study (7496%) [22]. The small
number of enrolled patients and the delay in
obtaining the 3D isotropic MR sequence af-
A B ter performing conventional MR arthrogra-
phy may be an explanation for the difference.
Sensitivities, specificities, and accuracies
for diagnosing anterior labral tears, subscap-
ularis tendon tears, and supraspinatus
infraspinatus tendon tears were comparable
to those of previous investigations [5, 23]
and showed no statistically significant dif-
ference between the two methods. For a le-
sion-to-lesion diagnosis of a supraspinatus
infraspinatus tendon tear, the sensitivity of
3D isotropic MR arthrography for a partial-
thickness articular surface tear was 50%
(3/6), which was relatively lower than the
sensitivity reported in previous studies with
3-T imaging [9] or imaging with abduction
C D
and an external rotation position [24]. The
Fig. 4Arthroscopically proven partial-thickness articular surface tear of supraspinatus tendon in 54-year-old
small number of tears may be responsible
man.
A and B, Fat-suppressed T1-weighted oblique coronal (A) and oblique sagittal (B) images show focal for this difference. The sensitivities and
accumulation of contrast material and indistinct margin at articular surface of supraspinatus tendon (arrows), specificities for a full-thickness tear and a
which was interpreted as partial-thickness articular surface tear of supraspinatusinfraspinatus tendon. partial-thickness bursal surface tear were
C and D, Three-dimensional isotropic MR arthrography oblique coronal (C) and sagittal (D) reformatted images
again show identical findings (arrows). Lesion was interpreted as partial-thickness articular surface tear of comparable to the sensitivities and specifici-
supraspinatusinfraspinatus tendon. ties reported in previous studies and showed

AJR:192, February 2009 477

 Oh et al.

raphy. AJR 1993; 161:12291235

2. Sommer T, Vahlensieck M, Wallny T, et al. Indi-
rect MR arthrography in the diagnosis of lesions
of the labrum glenoidale [in German]. Rofo 1997;
167:4651
3. Maurer J, Rudolph J, Lorenz M, et al. A prospec-
tive study on the detection of lesions of the labrum
Downloaded from www.ajronline.org by 202.159.8.37 on 02/18/17 from IP address 202.159.8.37. Copyright ARRS. For personal use only; all rights reserved

glenoidale by indirect MR arthrography of the

shoulder [in German]. Rofo 1999; 171:307312
4. Lee JH, Van Raalte V, Malian V. Diagnosis of
SLAP lesions with Grashey-view arthrography.
Skeletal Radiol 2003; 32:388395
5. Yagci B, Manisali M, Yilmaz E, et al. Indirect MR
arthrography of the shoulder in detection of rotator
cuff ruptures. Eur Radiol 2001; 11:258262
A B
6. Gold GE, Suh B, Sawyer-Glover A, Beaulieu C.
Fig. 5Arthroscopically proven calcific tendinosis with no tear of supraspinatus tendon in 57-year-old woman. Musculoskeletal MRI at 3.0 T: initial clinical ex-
A and B, Fat-suppressed T1-weighted oblique coronal (A) and 3D isotropic MR arthrography oblique coronal
reformatted (B) images show accumulation of contrast material at articular side of supraspinatus tendon perience. AJR 2004; 183:14791486
(arrows), which was interpreted as articular side partial-thickness supraspinatusinfraspinatus tendon tear. 7. Gold GE, Busse RF, Beehler C, et al. Isotropic
MRI of the knee with 3D fast spin-echo extended
no statistically significant differences be- obtained, followed by 2D conventional MR echo-train acquisition (XETA): initial experience.
tween the two methods. arthrography 30 minutes after IV contrast AJR 2007; 188:12871293
Although the imaging time for the 3D iso- injection. This procedure might cause an un- 8. Magee TH, Williams D. Sensitivity and specifici-
tropic MR arthrography sequence was much derestimation of the diagnostic value of 3D ty in detection of labral tears with 3.0-T MRI of
shorter than that of conventional MR arth isotropic MR arthrography in the diagnosis the shoulder. AJR 2006; 187:14481452
rography sequences, MPR required addition- of a shoulder lesion. Fourth, we evaluated 9. Magee T, Williams D. 3.0-T MRI of the supraspi-
al time. In this study, we were able to mini- images by a consensus instead of indepen- natus tendon. AJR 2006; 187:881886
mize the time for MPR by using a PACS- dent reviewer observation. Thus, we were not 10. Magee T. Can isotropic fast gradient-echo imag-
embedded program that supplied the capa- able to obtain interobserver agreement. Fi- ing be substituted for conventional T1-weighted
bility of simultaneous MPR. If MPR was to nally, we did not compare the image quality sequences in shoulder MR arthrography at 3 Tes-
be performed at another console by a tech- of an axial source image of 3D isotropic MR la? J Magn Reson Imaging 2007; 26:118122
nician or another radiologist, the advantages arthrography and the reformatted image. In 11. Legan JM, Burkhard TK, Goff WB 2nd, et al.
of the 3D isotropic MR arthrography se- theory, however, the quality of reformatted Tears of the glenoid labrum: MR imaging of 88
quence, including timesaving and the ability images is comparable to that of axial source arthroscopically confirmed cases. Radiology
to reconstruct in an arbitrary plane, would images on a 3D isotropic MR arthrography 1991; 179:241246
be decreased. sequence because 3D isotropic MR arthrog- 12. Probyn LJ, White LM, Salonen DC, Tomlinson G,
This study has several limitations. First, raphy consists of isotropic voxels. Boynton EL. Recurrent symptoms after shoulder
patient selection bias may have been intro- In conclusion, 3D isotropic MR arthrogra- instability repair: direct MR arthrographic as-
duced and resulted in the overestimation of phy with MPR is a method for the diagnosis sessmentcorrelation with second-look surgical
diagnostic performance because we enrolled of labral and rotator cuff lesions that is com- evaluation. Radiology 2007; 245:814823
in our study only patients who underwent ar- parable to conventional MR arthrography 13. Waldt S, Burkart A, Lange P, Imhoff AB, Rum-
throscopic shoulder surgery. Second, although but has a shorter imaging time. meny EJ, Woertler K. Diagnostic performance of
arthroscopy was regarded as the reference MR arthrography in the assessment of superior
standard for this study, it is an operator-de- References labral anteroposterior lesions of the shoulder. AJR
pendent method. Thus, misinterpretation dur 1. Chandnani VP, Yeager TD, DeBerardino T, et al. 2004; 182:12711278
ing arthroscopy was a probable source of er- Glenoid labral tears: prospective evaluation with 14. Kwak SM, Brown RR, Resnick D, Trudell D,
ror. Third, 3D isotropic MR images were MRI imaging, MR arthrography, and CT arthrog- Applegate GR, Haghighi P. Anatomy, anatomic

TABLE 2: Diagnostic Values of Conventional MR Arthrography (Method A) and 3D Isotropic MR Arthrography

(Method B) for Labral Abnormalities and Rotator Cuff Tears
Superior Labrum Anterior Labrum Subscapularis Tendon SSPISP Tendon
Method Method A Method B p Method A Method B p Method A Method B p Method A Method B p
Sensitivity 74% 70% 0.65 88% 100% 67% 67% 1.00 96% 96%
Specificity 54% 85% 0.10 96% 100% 85% 85% 1.00 75% 67% 0.56
Accuracy 67% 75% 0.36 94% 100% 75% 75% 1.00 89% 83% 0.56
NoteDash () indicates p value could not be calculated. SSPISP = supraspinatusinfraspinatus.

478 AJR:192, February 2009

 Conventional Versus Isotropic MR Arthrography of the Shoulder

variations, and pathology of the 11- to 3-oclock
position of the glenoid labrum: findings on MR
arthrography and anatomic sections. AJR 1998;
171:235238
15. De Maeseneer M, Van Roy F, Lenchik L, et al. CT
and MR arthrography of the normal and patho-
logic anterosuperior labrum and labralbicipital
Downloaded from www.ajronline.org by 202.159.8.37 on 02/18/17 from IP address 202.159.8.37. Copyright ARRS. For personal use only; all rights reserved
18. Totterman SM, Miller R, Wasserman B, Blebea
JS, Rubens DJ. Intrinsic and extrinsic carpal
ligaments: evaluation by three-dimensional Fou-
rier transform MR imaging. AJR 1993; 160: 117
123
19. Totterman SM, Miller RJ. Scapholunate ligament:
normal MR appearance on three-dimensional
MRI of the ankle with parallel imaging using
GRAPPA at 3 T. AJR 2007; 189:240245
22. Dinauer PA, Flemming DJ, Murphy KP, Doukas
WC. Diagnosis of superior labral lesions: com-
parison of noncontrast MRI with indirect MR
arthrography in unexercised shoulders. Skeletal
Radiol 2007; 36:195202

complex. RadioGraphics 2000; 20[spec no]:S67
S81
16. Pfirrmann CW, Zanetti M, Weishaupt D, Gerber
C, Hodler J. Subscapularis tendon tears: detection
and grading at MR arthrography. Radiology 1999;
213:709714
17. Kassarjian A, Bencardino JT, Palmer WE. MR
imaging of the rotator cuff. Radiol Clin North Am
2006; 44:503523, viiviii
gradient-recalled-echo images. Radiology 1996;
200:237241
20. Totterman SM, Miller RJ, McCance SE, Meyers
SP. Lesions of the triangular fibrocartilage com-
plex: MR findings with a three-dimensional gra-
dient-recalled-echo sequence. Radiology 1996;
199:227232
21. Bauer JS, Banerjee S, Henning TD, Krug R, Ma-
jumdar S, Link TM. Fast high-spatial-resolution
23. Rudolph J, Lorenz M, Schroder R, Sudkamp NP,
Felix R, Maurer J. Indirect MR arthrography in
the diagnosis of rotator cuff lesions [in German].
Rofo 2000; 172:686691
24. Herold T, Bachthaler M, Hamer OW, et al. Indi-
rect MR arthrography of the shoulder: use of ab-
duction and external rotation to detect full- and
partial-thickness tears of the supraspinatus ten-
don. Radiology 2006; 240:152160

F O R YO U R I N F O R M AT I O N
The comprehensive book based on the ARRS 2008 annual meeting categorical course on State-of-the-Art
Emergency and Trauma Radiology is now available! For more information or to purchase a copy,
see www.arrs.org.

AJR:192, February 2009 479