Documente Academic
Documente Profesional
Documente Cultură
Original Research
January 2009 to May 2011. Patients with prior were placed in the supine position, with the arm along
shoulder surgery were excluded. The patient group the body and the shoulder in a neutral position. Spi-
included 46 men and 10 women. Forty-one patients ral CT was only targeted to the affected shoulder, and
had recurrent shoulder anterior dislocation and the the anatomical area ranged from the acromioclavicu-
remaining 15 manifested shoulder pain and motion lar joint to the axillary recess. The CT parameters
limitation. The time between MR arthrography and used were: tube voltage 120 kV; tube current 100
MSCT ranged from 0–27 days (mean, 5.55 6 6.36 mAs; collimation beam 0.75 mm, feed/rotation 9 mm,
days). This project was approved by our Investiga- effective pitch 0.75. The FOV at acquisition was
tional Review Board of our hospital and informed con- 30 cm (pixel size, 0.6 0.6 mm) and section thick-
sent was obtained from all study patients. ness was 1 mm, with a section increment of 0.7 mm
(30% section overlap).
MRI
Image Analysis
All patients underwent MRI of the affected shoulder
using a 3.0 Tesla MR (Magnetom Trio with TIM sys- Both fat-suppressed 3D VIBE MR arthrography data
tem, Siemens, Erlangen, Germany) with a maximum and MSCT spiral data were then reformatted into par-
gradient amplitude of 44 mT/m and a maximum slew allel transverse, coronal oblique, and sagittal oblique
rate of 200 mT/m/s. Joint puncture was first per- images (section thickness, 3 mm; no intersection gap).
formed with fluoroscopic guidance via the rotator Transverse reformation was perpendicular to the gle-
interval approach, and then 10–12 mL of diluted nohumeral joint space. Coronal oblique reformation
gadopentetate dimeglumine with a concentration of 4– was parallel to the long axis of the glenoid. Sagittal
6 mmolGd/L was injected into the shoulder. The con- oblique reformation was not only perpendicular to the
trast material was prepared with a 20-mL syringe long axis of the glenoid, but also parallel to the gleno-
using 5 mL diluted Magnevist solution (Schering, humeral joint space on coronal oblique images, so
GuangZhou, China) with normal saline, 5 mL iodine- images en face to the glenoid articular surface could
based contrast (Omnipaque 300 mg I/mL; AnSheng, be collected.
ShangHai, China), and 5 mL lidocaine hydrochloride Two observers (both with more than 13 years of
injection (10). experience in musculoskeletal radiology) reviewed all
All patients were scheduled for MRI no more than reformatted images of fat-suppressed 3D VIBE MR
30 minutes after injection. A dedicated flexible surface sequences first, and then images of MSCT 2 weeks
coil (4-channel flex coil, large or small) supplied by later. The observers were blinded to patient identity
the MR manufacturer was used to receive the signal and clinical history.
from the affected shoulder. T1-weighted turbo spin- Signal-to-noise ratio (SNR) of the bone marrow and
echo images were routinely obtained in the transverse contrast-to-noise ratios (CNRs) between the bone mar-
plane (repetition time/echo time [TR/TE] 700/12 row and surrounding soft tissues with fat-suppressed
msec, 3.5 mm section thickness, 160 160 mm field 3D VIBE were assessed by one observer. Regions of
of view [FOV], matrix 256 256, 2 turbo factors), interest (ROIs) were placed within the glenoid bone
in the coronal oblique plane (TR/TE 700/12 msec, marrow, muscle belly of the infraspinatus, anterior
3.5 mm section thickness, 160 160 mm FOV, labrum, and intraarticular fluid on the transverse
matrix 256 256), and in the sagittal oblique plane reformatted images. The normalization to the noise
(TR/TE 650/11 msec, 4 mm section thickness, 160 was calculated by: S/N, where S corresponded to the
160 mm FOV, matrix 256 256). The coronal oblique mean signal intensity of the selected ROI and N corre-
images were arranged parallel to the long axis of the sponded to the standard deviation (SD) of the noise.
supraspinatus tendon and the sagittal oblique images The SD of the noise was obtained by placing an ROI
were perpendicular to the long axis of the supraspina- in a background area that was free of signal and as
tus tendon. large as possible. The CNRs were calculated by: (S1
In addition to the above routine clinical shoulder S2)/N, where S1 and S2 corresponds to the signal in-
MR, an arthrography-specific imaging protocol using tensity of the compared structures and N corresponds
a commercially available fat-suppressed 3D VIBE to the SD of the noise.
sequence was performed in each patient. The parame- The presence of a bony Bankart lesion was deter-
ters of the VIBE sequence were: TR/TE 12.2/4.9 mined on transverse reformatted images through the
msec, flip angle 10 , 160 160 mm FOV, matrix 512 glenoid. Bony Bankart lesions were diagnosed by
512, one slab of 112 slices with slice thickness of observing a bone fragment adjacent to the glenoid
0.6 mm, and one acquisition. This resulted in a voxel rim. In this analysis the two observers were working
volume of 0.6 0.6 0.6 mm. This isotropic 3D independently.
sequence was obtained in the coronal oblique plane The presence of glenoid bone loss was determined
and the acquisition time was 3 minutes 14 seconds. on sagittal oblique images en face to the glenoid fossa.
The best-fitting circle technique was used. Normally,
MSCT Imaging the shape of the inferior part of the bony glenoid in
the en face image can be circled by a best-fitting circle
All patients underwent nonarthrographic MSCT of the (3,11), and the anterior margin of the normal glenoid
affected shoulder on a 16-row multidetector CT unit fossa has a curved contour. Glenoid bone loss was
(Sensation 16, Siemens Medical Solutions). Patients defined as positive when that circle was missing, or
3D VIBE for Glenoid Bone Lesions 3
when the curved anterior margin of the glenoid fossa Bony Bankart lesions were noted in 23 (41.1%) of
became straight. 56 glenohumeral joints at nonarthrographic MSCT by
Quantification of glenoid bone loss was also based both observers. These 23 patients all had a history of
on sagittal oblique images en face to the glenoid fossa. recurrent anterior dislocation. Based on the results of
As stated above, with the best-fitting circle technique, MSCT, the sensitivity and specificity of fat-suppressed
the missing part of that circle could represent the size 3D VIBE for bony Bankart lesions were 95.7% (22/
of the glenoid bone loss. The percentage of glenoid 23) and 93.9% (31/33), respectively for observer 1,
bone loss could be calculated as the difference in gle- and 100% (23/23) and 97.0% (32/33), respectively for
noid defect width divided by the diameter of the circle observer 2 (Fig. 2–5). At interobserver comparison,
100% (Fig. 1). Detection and quantification of gle- agreement was very good (kappa value ¼ 0.926).
noid bone loss were performed by the same observer. Glenoid bone loss was found in 40 (71.4%) of 56
patients on nonarthrographic MSCT examination,
including 35 with recurrent anterior dislocation and
Statistical Analysis five with shoulder pain and motion limitation. Based
By using the results of nonarthrographic MSCT as the on the results of MSCT, the sensitivity and specificity
reference standard, the sensitivity and specificity of of fat-suppressed 3D VIBE in predicting glenoid bone
fat-suppressed 3D VIBE for qualifying bony Bankart loss were 95.0% (38/40) and 93.8% (15/16), respec-
lesions were determined. Cohen’s kappa coefficient tively. There were two false-negative fat-suppressed
was calculated to quantify the level of agreement at 3D VIBE assessments. In these two cases, nonarthro-
interobserver comparison. graphic MSCT showed 3.72% and 6.03% glenoid bone
Based on the results of MSCT, the sensitivity and loss, respectively. There was one false-positive fat-
specificity of a fat-suppressed 3D VIBE for qualifying suppressed 3D VIBE assessment. In this case, fat-
glenoid bone losses were determined. Additionally, the suppressed 3D VIBE revealed 6.69% glenoid bone
statistical significance of the percentages of glenoid
bone loss between fat-suppressed 3D VIBE MR
arthrography and nonarthrographic MSCT was calcu-
lated using the double sided paired-samples t-test.
Correlation between percentages of glenoid bone loss
with fat-suppressed 3D VIBE and MSCT was assessed
with Spearman’s rank coefficient (r).
Data entry procedures and statistical analyses were
performed with SPSS 11.5 software (Chicago, IL). P
0.05 was considered statistically significant.
RESULTS
The SNR of the bone marrow with fat-suppressed 3D
VIBE was 23.89 6 16.10. Bone marrow-to-anterior Figure 2. Bony Bankart lesion in a 21-year-old man. The
labrum CNR, bone marrow-to-infraspinatus CNR, axial reformatted image of fat-suppressed 3D VIBE (a) shows
and bone marrow-to-intraarticular fluid CNR were a low signal bony fragment attaching to the anteroinferior
40.739 6 23.572, 48.477 6 24.566, and 169.675 glenoid neck (arrow), which is demonstrated a bony Bankart
6 87.024, respectively. lesion (arrow) on the reformatted CT image (b).
4 Tian et al.
Figure 3. Bony Bankart lesion in a 27-year-old man. The Figure 5. False bony Bankart lesion with fat-suppressed 3D
axial reformatted image of fat-suppressed 3D VIBE (a) shows VIBE in a 29-year-old man. The axial reformatted image of
a low signal bony fragment adjacent to the anteroinferior gle- fat-suppressed 3D VIBE (a) shows a triangular low signal
noid neck (arrow), which is demonstrated as a detached structure attaching to the anteroinferior glenoid neck
bony Bankart lesion (arrow) on the reformatted CT image (b). (arrow), which was mistaken for a bony Bankart lesion by
both observers. This structure proved to be adipose tissue
adjacent to the glenoid neck (arrow) on the axial reformatted
loss, but no bone loss was apparent on nonarthro-
MSCT image (b).
graphic MSCT.
As to the quantification of glenoid bone loss,
fat-suppressed 3D VIBE MR arthrography (10.48% 6 lesions. By adding an additional fat-suppressed 3D
8.71%; range, 0%–30.21%) and nonarthrographic VIBE sequence, which is commercially available,
MSCT (10.96% 6 9.00%; range, 0%–31.40%) showed we can reliably evaluate glenoid bone abnormalities,
no statistical difference (t ¼ 1.072, P ¼ 0.288), and eliminating the need of MSCT examination for detecting
there was a high correlation (r ¼ 0.921, P < 0.001) bony abnormalities. The patient then can not only
between these two techniques (Fig. 6). The maximum avoid the radiation hazard, but also reduce the
difference in the quantification of glenoid bone loss total cost of the examination by eliminating the
was 7.43% (Fig. 7), with SD 3.33%. additional CT.
Fat-suppressed 3D VIBE is an interpolated T1-
weighted gradient-echo MR sequence with the fast low
DISCUSSION angle shot technique (12). Compared with the conven-
tional fat-suppressed 3D T1-weighted fast low angle
Our study demonstrates a high consistency between shot, the fat-suppressed 3D VIBE sequence has an
fat-suppressed 3D VIBE MR arthrography and nonar- ability to achieve images with higher spatial resolution
thrographic MSCT in detecting bony Bankart lesions in much shorter acquisition times (13); therefore, it
and glenoid bone loss and quantifying the glenoid bone
loss. Thus, the fat-suppressed 3D VIBE sequence in
MR arthrography may be a reliable replacement for
nonarthrographic MSCT for detecting glenoid bony
abnormalities. In patients with anterior shoulder dis-
location, MR arthrography routine sequences (SE or
FSE sequences) can be used for detecting labral
3. Sugaya H, Moriishi J, Dohi M, Kon Y, Tsuchiya A. Glenoid rim 13. Kataoka M, Ueda H, Koyama T, et al. Contrast-enhanced
morphology in recurrent anterior glenohumeral instability. volumetric interpolated breath-hold examination compared with
J Bone Joint Surg Am 2003;85:878–884. spin-echo T1-weighted imaging of head and neck tumors. AJR
4. Boileau P, Villalba M, Hery JY, Balg F, Ahrens P, Neyton L. Risk Am J Roentgenol 2005;184:313–319.
factors for recurrence of shoulder instability after arthroscopic 14. Vogt FM, Antoch G, Hunold P, et al. Parallel acquisition techni-
Bankart repair. J Bone Joint Surg Am 2006;88:1755–1763. ques for accelerated volumetric interpolated breath-hold exami-
5. Griffith JF, Yung PSH, Antonio GE, Tsang PH, Ahuja AT, Chan nation magnetic resonance imaging of the upper abdomen:
KM. MSCT compared with arthroscopy in quantifying glenoid assessment of image quality and lesion conspicuity. J Magn
bone loss. AJR Am J Roentgenol 2007;189:1490–1493. Reson Imaging 2005;21:376–382.
6. Waldt S, Burkart A, Lange P, Imhoff AB, Rummeny EJ, Woertler 15. Lauenstein TC, Goehde SC, Herborn CU, et al. Three dimensional
K. Diagnostic performance of MR arthrography in the assessment volumetric interpolated breath-hold MR imaging for whole-body
of superior labral anteroposterior lesions of the shoulder. AJR Am tumor staging in less than 15 minutes: a feasibility study. AJR
J Roentgenol 2004;182:1271–1278. Am J Roentgenol 2002;179:444–449.
7. Waldt S, Burkart A, Imhoff AB, Bruegel M, Rummeny EJ, Woer- 16. Mermuys KP, Vanhoenacker PK, Chappel P, Van Hoe L. Three
tler K. Anterior shoulder instability: accuracy of MR arthrography dimensional venography of the brain with a volumetric interpo-
in the classification of anteroinferior labroligamentous injuries. lated sequence. Radiology 2005;234:901–908.
Radiology 2005;237:578–583. 17. Chuang TY, Adams CR, Burkhart SS. Use of preoperative three-
8. Tung GA, Hou DD. MR arthrography of the posterior labrocapsular dimensional computed tomography to quantify glenoid bone loss
complex: relationship with glenohumeral joint alignment and clini- in shoulder instability. Arthroscopy 2008;24:376–382.
cal posterior instability. AJR Am J Roentgenol 2003;180:369–375. 18. d’Elia G, Di Giacomo A, D’Alessandro P, Cirillo LC. Traumatic an-
9. Bitzer M, Nasko M, Krackhardt T, et al. Direct MSCT-arthrography terior glenohumeral instability: quantification of glenoid bone
versus direct MR-arthrography in chronic shoulder instability: loss by spiral CT. Radiol Med 2008;113:496–503.
comparison of modalities after the introduction of multidetector- 19. Huijsmans PE, Haen PS, Kidd M, Dhert WJ, van der Hulst VP,
MSCT technology. Rofo 2004;176:1770–1775. Willems WJ. Quantification of a glenoid defect with three-dimen-
10. Shang Y, Zheng ZZ, Li X. Direct MR arthrography of the shoulder sional computed tomography and magnetic resonance imaging: a
at 3 Tesla: optimization of gadolinium concentration. J Magn cadaveric study. J Shoulder Elbow Surg 2007;16:803–809.
Reson Imaging 2009;30:229–235. 20. Saito H, Itoi E, Sugaya H, Minagawa H, Yamamoto N, Tuoheti Y.
11. Burkart SS, DeBeer JF, Tehrany AM, Parten PM. Quantifying gle- Location of the glenoid defect in shoulders with recurrent anterior
noid bone loss arthroscopically in shoulder instability. Arthro- dislocation. Am J Sports Med 2005;33:889–893.
scopy 2002;18:488–491. 21. Kralinger F, Aigner F, Longato S, Rieger M, Wambacher M. Is the
12. Zheng ZZ, Shan H, Li X. Fat-suppressed 3D T1-weighted gradient- bare spot a consistent landmark for shoulder arthroscopy? A
echo imaging of the cartilage with a volumetric interpolated breath- study of 20 embalmed glenoids with 3-dimensional computed
hold examination. AJR Am J Roentgenol 2010;194:W414–419. tomographic reconstruction. Arthroscopy 2006;22:428–432.