821

MR Diagnosis of Recurrent Tears

Downloaded from www.ajronline.org by 212.118.37.238 on 12/30/23 from IP address 212.118.37.238. Copyright ARRS. For personal use only; all rights reserved

in the Knee: Value of lntraarticular

Contrast Material

Gregory R. Appiegate1 OBJECTIVE. After surgical resection or repair of a torn meniscus, the healed area
Bonnie D. Flannigan1 may have areas of abnormal signal Intensity on MR images. Consequently, routine
Brad S. bun2 MR Imaging Is not reliable for detecting recurrent meniscal tears. As a result, we
studied the efticacy of MR imaging with Intraartlcular contrast material (MR arthrogra-
James M. Fox3
phy) for detecting recurrent tears of the meniscus.
Wilson Del Pizzo3 SUBJECTS AND METHODS. Thirty-seven patients who previously had a meniscal
tear treated by either meniscab resection or repair had conventional MR Imaging and
MR arthrography with 40-50 ml of a 1 :100 solution of gadopentetate dimegiumine in
saline. All patients had arthroscopy shortly after the MR studies. Follow-up arthro-
scopic surgery was performed within an average of 6.6 weeks after the MR arthro-
grams. The routine MR Images and MR arthrograms were reviewed separately and
randomly, and these results were compared with the arthroscoplc findings. Menlscal
morphology, signal Intensity, and the presence of joint fluid tracking into recurrent
tears were evaluated.
RESULTS. The overall accuracy in diagnosing recurrent menlscal tears In the post-
operative meniscus was 66% when conventional MR Imaging was used and 88%
when MR arthrography was used. In patients who had only minimal meniscal resec-
tion, both methods had an accuracy of 89%. In patients who had more extensive
menlscai resection, accuracy was 65% wIth conventional MR Imaging and 87% with
MR arthrography. In four patients who had only a small meniscal remnant, the accu-
racy was 50% with routine MR Imaging and 100% with MR arthrography. On conven-
tlonal MR images, the presence of an effusion tracking Into a meniscal tear had a
sensitivity and positive predictive value of 90% for detection of recurrent meniscal
tears; however, the sensitivity was only 41%.
CONCLUSION. Our results show that the sensitivity of MR imaging In detecting
menlscai tears after surgery varies with the extent of the resection. Sensitivity was
considerably improved when intraarticular contrast material was used. MR arthrogra-
phy should be considered as an alternative to arthroscopy in patients who have had
resection or repair of the meniscus.

AJR 1993;161 :821-825

MR imaging is valuable in the diagnosis of ligamentous and meniscal injuries of

the knee. It is not reliable, however, in the detection
of recurrent tears in menisci
that have been nesected on repaired surgically [1-4]. The appearance of the
Aeceived March 22, 1993; accepted after revi-
sion June 7, 1993.
meniscus after resection or repair is highly variable and depends on the location
1 Valley Presbyterian Magnetic Aesonance Cen- and size of the original tear and on the degree of resection or repair. The principal
ter, 6815 Noble Ave., Suite 105, Van Nuys, CA role of MR imaging in patients who have persistent and recurrent symptoms after
91405. Address correspondence to G. R. Apple- meniscal resection on repair is to differentiate recurrent meniscal tears from other
gate.
causes of pain.
2San Antonio Orthopedic Group, 1303 McCul-
The purpose of our study was to determine the sensitivity of MR imaging with
lough, #135, San Antonio, TX 78284.
intraarticular contrast material for detecting recurrent meniscal tears after sun-
3Southern California Orthopedic Institute, 6815
Noble Ave., Van Nuys, CA 91405. gary. MR arthrograms and conventional MR images were compared; meniscal
0361-803)(/93/1 61 4-0821
morphology, signal intensity, and the presence of joint fluid tracking into the tear
© American Roentgen Ray Society were evaluated.
 822 APPLEGATE ET AL. AJR:161 , October1993

Subjects and Methods ened a sign of a recurrent tear. Conventional T2-weighted sagittal
images were examined for a relative increase in intrameniscal sig-
Prospectively, conventional MR images and MR arthrognams nal intensity. Intrameniscal signal intensity equal to that of joint fluid
were obtained in 37 patients who previously had surgery for repair was thought to represent joint fluid tracking into a torn meniscal
of a torn meniscus. All patients had persistent on recurrent knee pain remnant. Joint fluid adjacent to an area of altered meniscal signal
after the surgery. Approval was obtained from the investigational
Downloaded from www.ajronline.org by 212.118.37.238 on 12/30/23 from IP address 212.118.37.238. Copyright ARRS. For personal use only; all rights reserved

intensity that did not track into the area was considered evidence of
review board at Valley Presbyterian Hospital, Van Nuys, CA. All a stable meniscal remnant despite the meniscal grade. The MR
patients gave informed consent. arthnognams were similarly graded for an increase in intrameniscal
The study group consisted of 19 women and 18 men 17-72 years signal intensity on the Ti -weighted MR arthrogram and for improved
old (mean, 40 years). Follow-up arthnoscopic surgery was pen- definition of the morphology of the meniscal remnant.
formed within an average of 6.6 weeks (range, 1.5-33 weeks) after The indications for arthroscopy in these patients included sus-
MR arthnognaphy. No patient had further injury during the time pected recurrent meniscal tears, neconstructions of the anterior cru-
between MR imaging and arthnoscopy. ciate ligament, removal of hardware, loose bodies, lateral
The MR images were obtained with a 1.5-T system (Signa, G.E. netinacular release, chondral lesions, and treatment of arthnofibnosis
Medical Systems, Milwaukee, WI) with a dedicated transmit-and- after previous reconstruction of the anterior cruciate ligament. The
receive extremity coil. Conventional MR images were of three types. arthnoscopic appearances of the menisci were divided into three
Spin-echo Ti -weighted (400-700/i 5-20 [TR/TE]) and T2-weighted groups on the basis of the extent of previous resection; the criteria
(2000-2200/20,80) images were obtained by using a 16-cm field of were the same as those used for MR grading. At arthnoscopy, a
view, 128 x 256 matrix, and one excitation. The section thickness meniscus was considered torn if any pattern of cleavage produced
was 3-5 mm for Ti-weighted images and 5 mm for T2-weighted an abnormality such that a portion of the meniscus could be dis-
images. For coronal spin-echo Ti -weighted images (400-700/15- placed from its normal position by probing. Arthnoscopy was the
20), a 14-cm field of view, i28-256 x 256 matrix, one excitation, and gold standard in this study.
section thickness of 3-5 mm were used.
For the MR arthrognams, 40-50 ml of a 1 :100 solution (2 mmol/l)
of gadopentetate dimeglumine (Magnevist, Beniex Laboratories, Results
Inc., Cedar Knolls, NJ) in saline was injected intnaarticulanly. This Included in the study were 29 medial and 1 2 lateral
was initially done with fluonoscopic guidance after sterile skin prepa-
menisci. Overall accuracy in the diagnosis of recurrent
ration, local administration of lidocaine, and a test injection of less
than 0.5 ml of Hypaque (Winthrop Pharmaceuticals, New York). A meniscal tears in the postoperative meniscus was 66% with
conventional 18-gauge needle was used. Currently, all patients are conventional MR imaging. Sensitivity was 69%, specificity
injected via a supralatenal approach on the MR table. Cane was was 60%, negative predictive value was 53%, and positive
taken not to introduce air into the joint and to drain the effusion predictive value was 75%. Use of intraarticular contrast
before injecting the contrast material. After injection and mild oxen- material improved the detection of recurrent tears and
cise of the knee, sagittal and coronal Ti -weighted MR arthrognams improved definition of meniscal morphology. With MR
were obtained. The parameters were the same as for the conven- arthrography, accuracy was 88%, sensitivity was 89%, spec-
tional images. Patients reported mild discomfort with the lidocaine ificity was 86%, negative predictive value was 80%, and pos-
administration. No complications were encountered. itive predictive value was 92% in the diagnosis of recurrent
All patients had arthnoscopy after the MR studies. Two expeni-
tear. Table 1 summarizes the results in comparison with
enced musculoskeletal radiologists reviewed the conventional MR
images and the MR arthrognams separately and randomly and arthnoscopic findings.
reached a consensus in cases of disagreement. They knew that the Diagnosis of recurrent tears according to the size of the
patient had had meniscal surgery but otherwise had no knowledge remaining meniscal remnant was analyzed (Table 2). Tears
of any clinical information. A surgeon recorded the findings at in the nine menisci with less than 25% nesected (Fig. 1)
arthnoscopy and reviewed the patient’s clinical history and previous could be evaluated accurately by using conventionally
surgical records without knowledge of the MR findings. The findings accepted MR criteria for determining a meniscal tear. Six
on conventional MR images and MR arthrograms were compared were retorn and three were stable at arthroscopy. In this
with surgical findings.
The menisci were divided anatomically into anterior, middle, and
posterior thirds, and an intrameniscal MR signal grade was TABLE 1 : Comparison of Conventional MR Imaging with MR
assigned to each part of the meniscus. The morphology of the Arthrography in the Detection of Recurrent Meniscal Tears
meniscus as seen on MR images was characterized according to After Surgery
the degree of meniscal resection into three groups: (A) less than
25% of the meniscus resected, (B) between 25% and 75% of the No. of Me nisci (n = 4i)
meniscus resected, and (C) more than 75% resected [4]. Any effu- Finding -___________ -----

Conventional MR Arthrognaphy
sion present was graded as small, moderate, or large. The presence
of abnormal signal in the meniscus was evaluated according to True-positive 18 24
accepted criteria [5]: grade I = an area with an irregular margin and False-positive 6 2
no communication with either articular surface, grade ii = a linear True-negative 9 12
area that does not reach an articulating surface, and grade ill = an False negative 8 3
area that extends to an articulating surface. If more than one grade Sensitivity (%) 69 89
of signal was observed in the meniscus, the highest grade was Specificity (%) 60 86
assigned. Grade Ill signal alone is known to be present in a stable Negative predictive value (%) 53 80
healed postoperative meniscus [i-3J, and therefore other features Positive predictive value (%) 72 92
were analyzed to establish meniscal stability. The presence of a Accuracy (%) 66 88
meniscal fragment separate from the residual meniscus was consid-
 AJA:161, October 1993 ENHANCED MR OF RECURRENT KNEE TEARS 823

TABLE 2: Diagnostic Accuracy Based on Size of Meniscal

1’
Remnant for Detection of Recurrent Meniscal Tears After
Surgery

No. of Me nisci (n = 41)

Group/Finding --------
Downloaded from www.ajronline.org by 212.118.37.238 on 12/30/23 from IP address 212.118.37.238. Copyright ARRS. For personal use only; all rights reserved

Conventional MR Arthrognaphy

0-25% Meniscal resection (n = 9)

True-positive 6 6
False-positive 1 1
True-negative 2 2
False-negative 0 0
-
Accuracy (%) 89 (71)a 89 (86)a
A B
25-75% Meniscal resection (n = 23)
True-positive 12 i5 Fig. 1 .-42-year-oid man with recurrent meniscal tear after mild resec-
False-positive 3 1 tion of posterior horn of lateral meniscus.
True-negative 3 5 A, Sagittab Ti-weighted conventional MR Image shows suspicious
grade Ill signal intensity within posterior horn (arrow) and blunting at free
False-negative 5 2 edge consistent with prior resection. T2-welghted Images (not shown)
Accuracy (%) 65 87 showed no increased Intensity.
> 75% Meniscal resection (n = 4) B, Sagittal MR arthrogram shows a truncated meniscal remnant
(arrowhead) with abnormal signal intensity extending to inferior articulat-
True-positive 0 1 ing surface representing a recurrent meniscal tear.
False-positive 2 0
True-negative 2 3
False-negative 0 0
Accuracy (%) 50 100
aOverall accuracy including five meniscal repairs/trephinations.

small group, the accuracy

MR imaging
was 89% with both conventional
and MR arthrography. in the two true-negative
[
cases, the menisci were characterized by blunted tips at the
resection site without any other abnormalities. The one false-
positive finding was in a lateral meniscus that had been
resected because of a tear involving the anterior and midpor-
tions of the meniscus with concomitant resection of a menis- A B
cal cyst. Conventional MR images showed a persistent grade
III signal in this area, which on MR arthnography appeared to Fig. 2.-48-year-old man with recurrent horizontal meniscal tear after
moderate menlscal resection.
enhance. At surgery, no recurrent tear was detected. A, Coronal short TRITE Ti-weighted conventional MR image shows a
For diagnosis of recurrent tears in menisci with 25-75% meniscal remnant (arrow) with grade II intrameniscal signal.
B, Coronal Ti -weighted MR arthrogram shows a recurrent tear (arrow)
nesected, the accuracy was 65% for conventional MR imag-
extending to the superior articulatIng surface.
ing and 87% for MR arthrognaphy. This group accounted for
the majority (23 menisci) of the postoperative menisci in this
study. Assessment based only on conventional criteria for
determining meniscal tears (i.e., grade III signal or irregular
morphology of the remaining meniscal remnant) was difficult
in this group (Fig. 2).
Four menisci had only a small meniscal remnant (>75%
nesected). In this limited group, findings on conventional MR
imaging were correct only twice; the findings were true-neg-
ative in both. On MR arthrograms, definition of the meniscal
remnant was improved, and recurrent tears could be cor-
rectly differentiated from stable meniscal rims (Fig. 3).
The presence of a joint effusion on T2-weighted conven-
tional images was thought to have a natural arthrographic
effect. We found effusions in 28 (85%) of the 33 patients in
whom T2-weighted images were obtained. The effusions
Fig. 3.-28-year-old man with a recurrent meniscal tear and small me-
were graded by size on the basis of the findings on conven- niscal fragment after extensive lateral menlscectomy.
tional MR images: 1 3 were small, 1 3 were moderate, and A, Coronal Ti-weighted conventional MR image shows a poorly de-
fined meniscal remnant suggestive of recurrent tear (arrow).
two were large. When an effusion was present, the overall B, Coronal MR arthrogram shows a well-outlined menbscal remnant
accuracy of T2-weighted images for detection of a recurrent (>75% resection). Also seen Is a small, subtle meniscal fragment (arrow).
 824 APPLEGATE ET AL. AJA:1 61, October1993

meniscal tear was only 56%. However, the specificity and had trephination of a peripheral partial tear of the inferior
positive predictive value of an effusion tracking into a tear surface of the medial meniscus. Follow-up MR imaging
(Fig. 4) for determining recurrent meniscal tears were both showed a residual grade II signal at the site of the tear. At
90%. Both the sensitivity and the negative predictive value MR arthnography, contrast material did not fill the undersur-
were 4i %, as shown by the presence of an effusion that did face defect at the meniscocapsular junction seen at arthros-
Downloaded from www.ajronline.org by 212.118.37.238 on 12/30/23 from IP address 212.118.37.238. Copyright ARRS. For personal use only; all rights reserved

not track into a recurrent tear, which was seen on MR arthro- copy. This may have been due to the fibrovascular scar
grams (Fig. 5). In this group, two diagnoses were faise-neg- overlying the incompletely healed tear, which could be
ative. Both menisci had residual grade Ill signal on detected only after careful probing of the meniscus.
conventional MR images but were thought to be stable
because of adjacent effusions that did not track into the
meniscus on the T2-weighted image. On MR arthrograms, Discussion
the contrast material tracked into the area in question, come- Although MR arthnography is gaining acceptance for use
lating with a tear. The remaining diagnosis that was false- in the shoulder [6, 7], its use in the knee has been limited.
negative on the basis of findings on both conventional MR Engel et al. [8] have improved the accuracy of MR imaging in
images and MR arthrograms was in a patient who previously the evaluation of cartilage in the knee by using MR arthrog-
naphy. A cadavenic study [9] of defects in articular cartilage
showed that MR arthrography had improved sensitivity in the
detection of meniscal tears, many of which were seen only
with MR arthrography. Data in human subjects, however, are
limited [8].
Subjectively, 32 of the 37 patients in our study were
improved after the more recent surgery. Five patients were
unchanged. Those patients who did not improve tended to
have had extensive meniscal resection (>75%) and associ-
ated degenerative changes within the joint. No patients in
this series were worse after arthroscopy.
Menisci in which resection or repair was minimal (<25%
resection) were easily evaluated on routine MR images, and
accuracy for detection of recurrent tears was acceptable.
These results are similar to those reported by others [i , 4].
Fig. 4.-6i -year-old woman with a large Joint effusion outlining the me- However, in menisci with more extensive resection (25-75%
niscal remnant and recurrent meniscal tear after moderate resection of
medial meniscus.
on >75% resected), detection was improved when intnaartic-
A, Ti-weIghted conventional MR image shows a “meniscal ghost” with ulan contrast material was used. The reason for this improve-
a poorly defined meniscal remnant (arrow) that Is difficult to distInguish ment, we think, is joint distension by contrast medium.
from adjacent cartilage.
B, Sagittab T2-weighted conventional MR image of knee with large ef- Increased volume and hydrostatic pressure within the joint
fusion shows natural arthrographic effect. Effusion fluid surrounds and improve delineation of the meniscal remnant encased by
defines menlscal remnant and recurrent tear (arrow). This case illus- contrast material, thereby improving evaluation of meniscal
trates that an effusion effectively outlinIng a recurrent tear can obviate
MR arthrography. morphology. Distension of the joint also separates the articu-
lar surfaces of the menisci from the articular cartilage, per-
mitting passage of contrast material into meniscal tears.
Although joint fluid was present in most of our patients, the
quantity was rarely sufficient to outline the meniscal remnant
completely, and therefore the joint fluid was less likely to
track into meniscal tears.
We also saw several poorly defined meniscal remnants on
routine MR images. With these “meniscal ghosts” (Fig. 4A),
it was difficult to determine where the resected meniscus
ended vs adjacent granulation tissue or fibrovascular scar.
The ghosts frequently had diffuse high signal intensity com-
pared with the signal intensity of the normal meniscus. The
meniscal remnants were adequately outlined on MR arthno-
A B grams, however.
By using a Ti shortening agent such as gadopentetate
Fig. 5.-31-year-old woman with a small recurrent meniscal tear after dimeglumine, high contrast-to-noise Ti -weighted images,
mild resection of medial meniscus.
which requires less time, could be obtained. Also, the con-
A, Sagittal T2-welghted conventional MR Image of knee with effusion
shows some blunting of free edge (arrow) In otherwise normal-appearing trast material tracking into the meniscal tears had quite high
meniscal remnant. signal intensity on Ti -weighted MR arthrograms. Recurrent
B, Sagittal MR arthrogram shows better definition of meniscal remnant
than A does. Note small tear (arrow) extending to the Inferior articulating
tears were easily differentiated from fibrovasculan scars and
surface. granulation tissues, which can be present in healed menis-
 AJR:161, October 1993 ENHANCED MR OF RECURRENT KNEE TEARS 825

cal tears and have lower signal intensity on short TR/TE Conclusions
sequences. Although injections of saline might produce ade-
We have described the problems encountered in evaluat-
quate joint distension and allow the tracking of fluid into
ing the meniscus in patients who have had surgery to treat
these meniscal tears, a T2-weighted sequence would be
meniscal tears. Conventional MR imaging is most accurate
required. The signal intensities of saline and of granulation
when meniscal resection has been minimal, yet it is less
Downloaded from www.ajronline.org by 212.118.37.238 on 12/30/23 from IP address 212.118.37.238. Copyright ARRS. For personal use only; all rights reserved

tissue/scarring may be quite similar on T2-weighted se-

accurate than the virgin meniscus. With further degrees of
quences, although further investigation into this may be nec-
resection, the accuracy of conventional MR imaging is
essary. Advances in fast spin-echo sequences with improved reduced considerably. Use of intraarticuiar contrast material,
contrast resolution, however, may improve the results with with distension of the joint, improves detection of recurrent
long TRITE sequences.
meniscal teams in all degrees of meniscal resection when
For this study, we used a i :i 00 dilution of gadopentetate
compared with conventional MR imaging. Although most
dimeglumine in saline (2 mmol/I). This dilution is the optimal
patients have joint effusions, the effusions frequently are
concentration, according to studies by Engel et al. [8]. This is insufficient to adequately outline the entire meniscal remnant,
higher than the concentration we used in studies of the
and in our experience recurrent meniscal tears were seen
shoulder [6]. We currently use this 1 :iOO dilution for both
infrequently. The presence ofjoint fluid tracking into a menis-
shoulder and knee studies. cal tear, however, was a reliable sign of recurrent tear and, if
The selection bias in this study is recognized. All the
present, can obviate MR arthrogmaphy. Although we did not
patients were symptomatic and thus likely to have intraartic-
investigate the use of MR arthrognaphy for delineation and
ular abnormalities as a source of their pain. Also, with the
detection of nonmeniscal causes of recurrent pain, the tech-
techniques used, the proton density-weighted and T2- nique is better than conventional MR imaging for detection of
weighted conventional MR images had a lower signal-to-
cartilaginous lesions, plicae, and osteochondnitis dissecans
noise ratio than did some of the Ti -weighted MR arthro-
in normal knees [9, i2, 13]. Our results suggest that MR
grams. Also, the parameters were altered during the course
arthrography is a reasonable alternative to repeat arthros-
of the study as imaging strategies evolved. Short TR
copy in patients who have had surgical treatment of menis-
sequences for conventional MR imaging and MR arthnogra-
cal tears. Further studies of larger populations with new
phy, however, were identical for each patient.
high-resolution sequences will determine the ultimate clinical
Gadopentetate dimegiumine has not been approved by
usefulness of this technique.
the Food and Drug Administration for intraarticuiar use.
Studies in both animals and humans [8, i 0] have concluded
that although gadopentetate dimeglumine is taken up by car- REFERENCES
tilage, this process is reversible. By i 7 hr, no contrast mate- 1 . Smith DK, Totty WG. The knee after partial meniscectomy: MR imaging
nial was found within viable human cartilage [8]. features. Radiology i990;176:141-144
If fluoroscopy is used, the total examination requires 90 2. Deutsch AL, Mink JH, Fox JM, et al. Peripheral meniscal tears: MR find-
ings after conservative treatment or arthroscopic repair. Radiology 1990;
mm. Currently, without fluonoscopy, the examination can be 176:485-488
complete in 1 hr. Forty-five minutes of scanning was 3. Farley TE, Howell SM, Love KF, et al. Meniscal tears: MR and arthro-
required at the time the study began. For throughput consid- graphic findings after arthroscopic repair. Radiology 1991180:517-522
erations, while one patient was at fluoroscopy, another 4. Deutsch AL Mink JH, Fox JM, et al. The postoperative knee. Magn Reson 0
i992;8:23-54
patient could be imaged. The cost ofthe added time required
5. Crues JV, Mink J, Levy T, et al. Meniscal tears of the knee: accuracy of
for this study was not recovered. In view of the high cost of MR imaging. Radiology i987;164:445-448
MR imaging, however, the diagnostic accuracy for detecting 6. Flannigan B, Kursunoghi-Brahme 5, Snyder SJ, et al. MA arthrography of
recurrent tears must be high. Cost effectiveness and assess- the shoulder: comparison with conventional MA imaging. AJR
i990;155:829-832
ment of patient outcome were not goals of this study. Our
7. Hodler J, Kursunoghi-Brahme 5, Snyder SJ, et al. Rotator cuff disease:
results, however, can be extrapolated and compared with assessment with MA arthrography versus standard MA imaging in 36
prior studies [ii] demonstrating the use of routine MR imag- patients with arthroscopic confirmation. Radiology 1 992;1 82:431-436
ing in the virgin knee, which allow for more appropriate 8. Engel A, Hajek P, Kramer J, et al. Magnetic resonance knee arthrography
selection of patients who would benefit from arthroscopy. enhanced contrast by gadolinium complex in the rabbit and in humans.
Acta Orthop Scand Suppl i990;61[240J:1-55
Similarly, in the postoperative meniscus, MR arthrography
9. Gylys-Morin VM, Hajek PC, Sartoris DJ, Aesnick D. Articular cartilage
may help establish the stability of meniscal remnant, evalu- defects: detectability in cadaver knees with MA. AJR i987;148:11 53-
ate the ligaments, and evaluate the cartilaginous surfaces. 1157
These evaluations may be used to more appropriately select 10. Hajek PC, Sartoris DJ, Gylys-Morin V, et al. The effect of intraarticular
gadolinium-DTPA on synovial membrane and cartilage. Invest Radiol
patients in whom arthroscopy could have therapeutic impli-
i990;25:1 79-183
cations. Our data and the results of others have shown that 11 . Auwe P, McCarthy 5, Wright J, Randall L, Lynch K, JokI P. Does MR
conventional MR imaging is not accurate for the detection of imaging effectively replace diagnostic arthroscopy? Radiology 1992;
recurrent tears in surgically repaired menisci. Therefore, the 183:335-339
added time and mildly invasive nature of MR arthrogmaphy in 12. Hajek PC, Sartoris DJ, Neumann CH, Resnick D. Potential contrast
agents for MR arthrography: in vitro evaluation and practical observa-
this select situation may be justifiable. Rapid spin-echo
tions. AJR i987;149:97
sequences and arthrocentesis on the MR table will increase 13. Hajek PC, Baker LL, Sartoris DJ, et al. MR arthrography: an anatomic-
the efficiency of this technique. pathologic investigation. Radiology i987;163:141 -147