Mogneric Resonance Printed in the USA.

Imaging. Vol. 10, pp. 531-539, All rights reserved.

1992 Copyright

0

0730-725X/92 $5.00 + .OQ 1992 Pergamon Press Ltd.

l Original Contribution

MR KNEE IMAGING: AXIAL 3DFT GRASS PULSE SEQUENCE VERSUS SPIN-ECHO IMAGING FOR DETECTING MENISCAL TEARS S. AUBEL,*

R.L.

HEYD,~

EL.

THAETE,*

AND P. WOZNEY*$

*University of Pittsburgh, Pittsburgh NMR Institute, Pittsburgh, PA 15213 USA tSt. Elizabeth Hospital Medical Center, Department of Radiology, Youngstown, OH 44501. The knees of 17 patients (34 menisci) referred for magnetic resonance (MR) imaging to evaluate knee pain were examined using thin axial three-dimensional Fourier transform QDFT) gradient-refocused acquisition in a steady state (GRASS) images through the menisci, to determine if this method is sensitive and specific for detecting meniscal tears. Results were compared with spin-echo images with long TR and double-echo TE in both coronal and sagittal planes. Arthroscopy results, available in each case, were used as the “gold standard.” Twelve meniscal tears were identified at arthroscopy. Axial 3DFT GRASS technique detected 10 of the 12 meniscal tears compared to 9 or 12 using spin-echo technique. With axial 3DFT GRASS technique one false-positive meniscal tear was reported, compared with two false-positive tears on spin-echo images. Axial 3DFT GRASS images were very useful in detecting peripheral tears, showing displaced meniscal fragments, and evaluating complex tears. In this small study, thin axial 3DFT GRASS images were comparable to spin-echo images for detecting meniscal tears, and were helpful in complicated cases in which they provided complementary information to that obtained from spinecho images.

Keywords: Menisci, Knee; 3D; Gradient echo; Pulse sequence. MR imaging has become a preferred method for evaluating knee pathology, and numerous MR scanning techniques have been investigated. While spin-echo sequences are generally used, the optimum scanning sequence is still evolving. Recent development of threedimensional Fourier transform (3DFT) gradient-refocused acquisition in a steady state (GRASS) technology allows contiguous images to be produced along with thin sections, which could improve the accuracy of MR studies by reducing volume averaging and interslice gaps. Furthermore, axial orientation, which provides a different view of meniscal anatomy than seen on standard sagittal and coronal images, has the potential to further improve accuracy of scan interpretation. Combining these two factors by producing axial 3DFT GRASS images was considered a possible way to improve knee imaging. The purpose of this study was to determine if thin axial 3DFT GRASS images of the knee are useful in RECEIVED 12/23/91;

detecting meniscal tears. Other types of knee pathology, such as bone or ligament abnormalities, were not evaluated. Axial 3DFT GRASS images were compared with sagittal and coronal spin-echo images and with arthroscopy results. SUBJECTS

AND METHODS

Seventeen patients, ranging in age from 14 to 59 years (mean age, 29 years), referred for evaluation of knee pain during a 4-mo period were studied with MR imaging. Arthroscopic confirmation was available on all patients. Arthroscopy was performed by six university faculty orthopedic surgeons specializing in sports medicine. In most cases arthroscopy was performed within 2 wk of MR imaging and in all cases arthroscopy was completed within 2 mo of the MR examination. MR imaging was performed on a General Electric Signa system (Milwaukee, WI) operating at 1.5 T. A

ACCEPTED 2/21/92.

$Present address: 33 E. Marion Ave., Punta Gorda, FL 33950. Paper presented at the 1990 Annual Meeting of the Atnerican Roentgen Ray Society, Washington, D.C.

Address correspondence to (present address): S. Aubel, M.D., St. Elizabeth Hospital Medical Center, Department of Radiology, 1044 Belmont Ave., PO Box 1790, Youngstown, OH 44501-1790. 531

532

Magnetic Resonance Imaging 0 Volume 10, Number 4, 1992

axial images, a deformity with free flaps, or a discontinuous edge with an absent meniscal segment.

dedicated transmit-and-receive extremity coil was used in each case. Patients were scanned in the supine position with the knees extended and externally rotated lo-15 degrees. Sagittal and coronal 2DFT spin-echo images were acquired using the following parameters: repetition time (TR) 2200 msec with echo delay (TE) 20 and 80 msec. Sagittal slices were 3.0 mm thick, with 1.O mm interslice gaps and two excitations. Coronal slices were 5.0 mm thick, with 1.O mm interslice gaps and two excitations. A 14-16 cm field of view (FOV) was used according to patient size. A 128 x 256 matrix was selected. Scanning time was approximately B-10 min. GRASS is the variable-flip-angle gradient-refocused pulse sequence available on the General Electric MR system. By using the 3DFT technique, axial GRASS images were obtained with the following parameters: TR 65-75 msec, TE 13-15 msec, nominal flip angle 25-30 degrees, slice thickness 0.7-l .O mm, and FOV 14-16 cm, according to patient size. Thirty-two contiguous 0.7-I .O mm thick slices were acquired using a 128 x 256 matrix and one excitation. Menisci were visualized on five to eight images. Scanning time for the axial 3DFT GRASS studies was approximately 5 min. On spin-echo images, menisci were assessed using criteria published by Reicher et al.’ and Stoller et a1.2 Horizontal areas of increased signal, not communicating with the surface, were considered degenerative changes. A meniscus was considered torn if a linear area of increased signal extending to the meniscal surface was observed. Axial 3DFT GRASS images were interpreted using the following criteria:

All images generated during both studies on each patient (using spin-echo and axial 3DFT GRASS technique) were combined and prospectively interpreted as a single examination by one of three MR-trained radiologists, two of whom (FLT and PW) participated in this study. This prospective interpretation is referred to as REPORT in this paper. The spin-echo and axial 3DFT GRASS images were then separated and retrospectively reviewed in random order as independent studies by three observers (SA, FLT, PW) “blinded” to results of the individual study not being analyzed and to arthroscopic findings. Attention was directed only to menisci which were evaluated for the presence of degenerative changes or tears. The data used for analysis specified only the presence of meniscal tear or degeneration independent of location. Final determinations were made by majority opinion of the observers. Arthroscopy results used as the “gold standard” for comparison in this study were obtained from operative arthroscopy reports. RESULTS Normal Anatomy 3DFT GRASS images of the knee provide good anatomic detail of the menisci. In Fig. 1 an axial 3DFT GRASS image is compared with anatomy demonstrated on a cadaver knee joint. The normal menisci, which are composed of fibrocartilage, appear as uniform rounded or semicircular structures with homogeneous decreased signal.

1. A meniscus was considered normal if it exhibited diffuse low signal without discontinuity or area of increased signal. 2. Degenerative changes were defined as globular or linear areas of increased signal on only one or two images, with normal low signal on contiguous images and with no extension to inner or outer meniscal edge. 3. A tear was defined by a bright line on two or more

Table 1. Number

of menisci

detected

Arthroscopy, MR sequence Prospective REPORT (all images) Retrospective axial 3DFT GRASS Retrospective spin-echo n = 34.

True negative 18 21 20

Axial 3DFT GRASS vs. Spin-Echo Imaging Results obtained for prospective REPORT on each patient, retrospective analysis of spin-echo images and retrospective analysis of axial 3DFT GRASS images are compared with arthroscopy results in Table 1. In the 17 knees examined using both axial 3DFT GRASS and spin-echo technique, 12 meniscal tears were con-

on

MR imaging versus knee arthroscopy

negative

(n = 22)

False positive 4 1 2

Arthroscopy, True positive 9 10 9

positive

(n = 22)

False negative 3 2 3

MR knee imaging 0 S. AUBELET

AL.

533

(A)

Fig. 1. (A) Axial view of cadaver knee and (B) 3DFT axial image through the menisci demonstrating both the medial (large black arrow) and lateral meniscus (long white arrow). Also demonstrated are the anterior (small black arrow) and posterior (short white arrow) cruciate ligaments, along with meniscal ligamentous attachments (open arrow).

firmed at surgery. During initial prospective REPORT, using both sets of images, 9 of these tears were identified. Ten meniscal tears were identified by blinded retrospective review of the axial 3DFT GRASS images alone, compared to 9 tears identified using only spin-echo images.

At initial REPORT there were four false-positive meniscal tears, compared with one for axial 3DFT GRASS images and two for spin-echo images. Calculated sensitivity, specificity, and accuracy rates for prospective REPORT and retrospective review of axial 3DFT GRASS and spin-echo images are

534

Magnetic Resonance Imaging 0 Volume 10, Number 4, 1992 Table 2. Sensitivity,

specificity,

and accuracy

Sensitivity (70)

MR sequence Prospective REPORT Retrospective axial 3DFT GRASS Retrospective spin-echo

75 83 75

rates of MR imaging

Specificity Vo)

versus arthroscopy Positive accuracy (70)

82 95 91

70 90 82

Negative accuracy @J) 86 91 87

n = 34.

Magnetic resonance imaging has become a valuable modality in evaluating knee pathology. Because it is well suited for imaging cartilage, MR is particularly useful for evaluating menisci. Multiple studies1,3-7 have demonstrated MR imaging to be both sensitive and specific for evaluating meniscal tears. Most commonly, spin-echo sequences are used for knee imaging. Most meniscal tears can be seen on sagittal spin-echo T,-weighted or proton-density images. The coronal plane is useful in identifying “bucket handle” and “parrot beak” tears.

Three-dimensional gradient-echo techniques can be used to generate ultrathin slices with high signal-tonoise ratios.8 Comparison of fast 3D gradient-echo sagittal and coronal imaging with arthroscopy on 20 patients’ showed that all arthroscopically diagnosed meniscal tears and high-grade degenerative changes were detected with fast 3D MR imaging. Reeder et al.,9 in a study of 50 patients, found no statistically significant differences in the sensitivity and specificity of gradient-echo and spin-echo sequences using sagittal and coronal images. Three-dimensional Fourier transform (3DFT) data acquisition provides several advantages compared to conventional spin-echo pulse sequences for knee imaging: 1) improved spatial resolution and signai-to-noise ratio for given slice thickness; 2) truly contiguous slices; and 3) the ability to obtain isotropic voxels en-

(A)

(B)

presented in Table 2. Axial images yielded sensitivity, specificity and accuracy rates comparable to those obtained using coronal and sagittal spin-echo images. DISCUSSION

Fig. 2. (A) Coronal spin-echo image demonstrating a horizontal tear of the lateral meniscus (arrow), and (B) corresponding axial image also demonstrating oblique horizontal tear at the midportion of the lateral meniscus, with inner edge discontinuity (arrowheads).

MR

knee imaging 0 S.

abling data reformation in nonorthogonal planes. lo Prolonged scanning time is the major disadvantage of 3DFT spin-echo data acquisition. Using limited-flipangle gradient refocused pulse sequences such as GRASS allows shortened repetition times to be used and high-resolution images to be generated in a short time. Spritzer et al. lo examined 18 knees using sagit-

535

AUBEL ET AL.

tal3DFT GRASS imaging and compared results with sagittal spin-echo MR imaging and arthroscopy. Both MR imaging techniques correctly identified seven of eight meniscal abnormalities. However, using sagittal 3DFT GRASS imaging resulted in fewer false positive

(4 (4

(B) (f-9 Fig. 3. (A) Sagittal spin-echo image demonstrating oblique tear of the posterior medial meniscus, and (B) corresponding axial image demonstrating better delineation of complex tear at middle and posterior aspects of medial meniscus (arrows).

Fig. 4. (A) Sagittal spin-echo meniscal images showing a normal variant in the anterior horn, interpreted as a lateral meniscus tear (arrow). (B) Corresonding axial image showing linear increased signal within midportion of lateral meniscus (arrowheads), but corresponding images both directly superior and inferior were normal and therefore degeneration was diagnosed. No tear was found at arthroscopy.

536

Magnetic Resonance Imaging 0 Volume 10, Number 4, 1992

men&al tears (three vs. six), suggesting that overall accuracy of MR knee imaging could be improved by using sagittal 3DFT GRASS technique. In this study, axial 3DFT GRASS images were found to effectively identify meniscal tears. Ten of 12 tears were correctly identified. In Fig. 2 a lateral me-

niscal tear is seen on the axial 3DFT GRASS image as a linear area of increased signal disrupting the free edge. The tear is confirmed on the corresponding coronal spin-echo image. On the axial 3DFT GRASS images, the extent and severity of meniscal tears were often better delineated than on coronal and sagittal images. This is seen in Fig. 3, in which the spin-echo sagittal image shows an oblique tear of the posterior medial meniscus, while the corresponding axial image demonstrates a much larger area of abnormality. Decreased volume averaging with normal structures, which improves resolution on axial images, is particularly helpful in evaluating complex tears and radial longitudinal tears. The axial images provided better visualization of peripheral tears along with meniscal capsular separations because the axial plane shows the meniscal edge better, and because thin axial slices were obtained. Meniscal fragments are usually displaced in the axial plane, allowing good visualization with the axial sequence. In some cases, axial imaging showing displaced meniscal fragments confirmed a suspected meniscal tear. Two false-positive results were identified on spin-

(A)

(4 03 Fig. 5. (A) Sagittal spin-echo meniscal images demonstrating an area interpreted as degeneration of the posterior medial meniscus (arrow). (B) Corresponding axial image showing definite disruption along the posterior medial meniscal capsular insertion (arrows), which was confirmed at arthroscopy.

Fig. 6. (A) Axial image demonstrating severe masceration of the entire lateral meniscus (arrows), along with an area of high signal at the posterior medial meniscus (arrowhead, which persisted on multiple images and was diagnosed as a tear. (B) Coronal spin-echo image confirming lateral meniscus masceration (arrows). (C) Sagittal spin-echo image demonstrating only degeneration within the posterior medial meniscus (arrow), which was confirmed at arthroscopy. (Fig-

ure continued on facing page. )

MR knee imaging 0 S. AUBEL ET AL.

09

(C) Fig. 6 continued.

537

538

Magnetic Resonance Imaging l .Volume 10, Number 4, 1992

echo images. In one of these cases (Fig. 4), the axial images demonstrated no meniscal tear. Sagittal spinecho images that showed normal variation in the anterior horn of the lateral meniscus were incorrectly interpreted as showing a tear, while the corresponding axial images showed only meniscal degeneration. Axial scans were also slightly more effective in eliminating false-negative diagnoses of meniscal tears as compared with spin-echo technique. Two tears were missed in analyzing axial 3DFT images, compared with three in analyzing spin-echo images. A false-negative spin-echo result is seen in Fig. 5. Standard spinecho sagittal images demonstrate a linear area in the posterior medial meniscus interpreted as a degenerative change. On the corresponding axial image, disruption of the posterior capsular insertion, which was confirmed at arthroscopy, is identified. There are several additional advantages of axial plane imaging for evaluating menisci. Interpretation time is decreased with fewer meniscal images (five to eight) to review rather than the multiple meniscal images (30-40) obtained using spin-echo technique. It is easier to concentrate on the smaller number of axial images. Yet, due to the high resolution of axial images, diagnostic accuracy is maintained. Axial plane images are similar to the view seen by the arthroscopic surgeon, increasing their usefulness to the surgeon. There are several pitfalls in evaluating spin-echo images in the coronal and sagittal plane.” These include normal structures that can mimic meniscal tears. The transverse ligament and the lateral inferior genicular artery can appear as tears of the anterior horns of the medial and lateral menisci. In addition, the popliteus tendon can be mistaken for a tear in the posterior horn of the lateral meniscus. Concavity at the outer edge of the meniscus can create a volume-averaging artifact, mimicking a horizontal meniscal tear. Axial plane images, which provide another view of the anatomy, can help differentiate normal structures from meniscal tears. Axial images, however, were not without their own false results. An example is shown in Fig. 6. An obvious mascerated tear of the lateral meniscus is shown on both the coronal spin-echo and axial 3DFT GRASS scans. The axial scan also demonstrates an abnormal area in the posterior medial meniscus persisting on multiple images and interpreted as a tear. The corresponding sagittal spin-echo scan demonstrates degenerative changes of the medial meniscus which were confirmed at arthroscopy. However, this is the only false-positive axial 3DFT GRASS interpretation in this study, and in this particular patient did not change the clinical outcome or produce unnecessary surgery because of the obvious contralateral meniscal damage.

In the axial plane, horizontal tears are difficult to differentiate from linear degeneration, because the plane of the tear parallels the plane of the axial images. Nonperipheral tears involving the superior or inferior aspects of the menisci may be difficult to detect on axial images due to partial volume averaging. While coronal and sagittal spin-echo images have proven useful for evaluating extra-meniscal tissues such as cruciate and collateral ligaments, this has not been proven for thin axial GRASS images. In analyzing the spin-echo, axial, and REPORT findings, both the spin-echo and axial 3DFT GRASS studies yielded improved diagnosis over the REPORT. This could be due to the fact that three readers simultaneously retrospectively analyzed the data versus multiple readers individually doing the prospective REPORT interpretation. Only menisci were evaluated during retrospective analysis while the entire knee, including bone, ligaments, soft tissue, and menisci was prospectively evaluated. In addition, retrospective review was performed in two intense sessions, while prospective reporting was done over a 4-mo period. This study is not large enough to define statistically-significant differences between imaging techniques. Larger studies would be required for this purpose. In conclusion, accuracy of meniscal tear detection by axial 3DFT GRASS imaging was comparable to sagittal and coronal spin-echo imaging in this small series. Peripheral tears and meniscal capsular separation along with free and displaced fragments were better seen on axial 3DFT GRASS images than on sagittal and coronal spin-echo images and, in many cases, extent of meniscal damage was more apparent. Larger studies are required to confirm these initial results. We believe that axial 3DFT GRASS images are useful for evaluating meniscal tears. Axial meniscal images could be generated as a separate imaging sequence or reconstructed from a 3DFT data set to supplement standard technique. Further investigation is needed to better define the role of axial plane images in MR evaluation of the menisci. REFERENCES Reicher, M.A.; Hartzman, S.; Duckwiler, G.R.; Bassett, L.W.; Anderson, L.J.; Gold, R.H. Men&al injuries: detection using MR imaging. Radiology59:753-757; 1986. Staller, D. W.; Martin, C.; Crues, J.V., III; Kaplan, L.; Mink, J.H. Meniscal tears: Pathologic correlation with MR imaging. Radiology 163:731-735; 1987. Mink, J.H.; Deutsch, A.L. Magnetic resonance imaging of the knee. Clin. Orthop. 244~29-46; 1989. Crues, J.V. III; Mink, J.; Levy, T.L.; Lotysch, M.; Stoller, D.W. Meniscal tears of the knee: Accuracy of MR imaging. Radiology 164:445-448; 1987.

MR

knee imaging 0 S.

5. Mink, J.H.; Levy, T.;Crues, J.V., III. Tears of the anterior cruciate ligament and menisci of the knee: MR imaging evaluation. Radiology 167:769-774; 1988. 6. Polly, D.W., Jr; Callaghan, J.J.; Sikes, R.A.; McCabe, J.M.; McMahon, K.; Savory, C.G. The accuracy of selective magnetic resonance imaging compared with findings of arthroscopy of the knee. J. Bone Joint Surg. 7OA:192-198; 1988. 7. Glashow, J.L.; Katz, R.; Schneider, M.; Scott, W.N. Double-blind assessment of the value of magnetic resonance imaging in the diagnosis of anterior cruciate and meniscal lesions. J. Bone Joint Surg. 71A:113-119; 1989.

AUBEL ET AL.

539

8. Tyrrell, R.L.; Gluckert, K.; Pathria, M.; Medic, M.T. Fast three-dimensional MR imaging of the knee: comparison with arthroscopy. Radiologv 166:865-872; 1988. 9. Reeder, J.D.; Matz, S.O.; Becker, L.; Andelman, SM. MR imaging of the knee in the sagittal projection: comparison of three-dimensional gradient-echo and spinecho sequences. AJR 153:537-540; 1989. 10. Spritzer, C.E.; Vogler, J.B.; Martinez, S., et al. MR imaging of the knee: preliminary results with a 3DFT GRASS pulse sequence. AJR 150:597-603; 1988. 11. Herman, L.J.; Beltran, J. Pitfalls in MR imaging of the knee. Radiology 167:775-781; 1988.

MR knee imaging: axial 3DFT GRASS pulse sequence versus spin-echo imaging for detecting meniscal tears.

The knees of 17 patients (34 menisci) referred for magnetic resonance (MR) imaging to evaluate knee pain were examined using thin axial three-dimensio...
4MB Sizes 0 Downloads 0 Views