MR Imaging of the Knee: Three-Dimensional Fourier Transform G RASS Technique 1

Transcription

1 심Journal of the Korean Radiological Society 1996: 34(4) : MR Imaging of the Knee: Three-Dimensional Fourier Transform G RASS Technique 1 Oong Joo Kim, M.O., Young-uk Lee, M.O., Eun Kyung Youn, M.O., In Gye No, M.O., Seoung Bum Chin, M.O., Joon Sik Kim, M.0.2, Jae Yeul Choi, M.0.2 Purpose: To evaluate the usefulness of three-dimensional(30) Fourier transform(ft) gradient refocused acquisition in steady state (GRASS) technique for MR imaging ofthe knee. Materials and Methods: Sixty-three knees in 61 patients were imaged on the 1. 5T MR system. Wecompared 30FTGRASStechniquewith 20spin echo(se) technique in terms of conspicuousness of the lesions of internal knee structures based on the results of arthroscopy or open surgery. As a SE technique, sagittal T1- and T2-weighted, and coronal fat-suppressed T2-weighted sequences were performed. Sixty contiguous axial scans with 0.7 or 1 mm section thickness were performed using 30 GRASS technique, and we also evaluated arbitrarily reformatted images produced from the original axial voxel images. Results: For the depiction of the tear, 30FT GRASS was superior to 20 SE in three cases of medial meniscus, one of lateral meniscus, and two of anterior cruciate ligament. Specificity of 30 GRASS was also higher than that of 20 SE in evaluation of lateral meniscus and anterior cruiciate ligament. There was no significant difference in MR diagnosis fortears ofthe posterior cruciate, medial collateral, and lateral collateralligaments. 30 GRASS wassuperior in evaluating the extent and morphology of the torn menisci. Conclusion: The 30FT GRASS technique was comparable or even superior to the 20 SE technique in the evaluation of the internal structure of the knee, and can be expected to supplement standard MR knee techniques, especially in complicated casesof meniscal or ligamentoustears. Index Words: Knee, MR Knee, i 미 unes Knee, ligaments, menisci, and cantilage Magnetic resonance(mr), p 비 se sequences INTRODUCTION Magnetic resonance (MR) imaging has become a preferred method for evaluating the internal derangement of the knee. The imaging technique is diverse. Two -dimensional(20) Fourier transform(ft) spin echo (SE) sequences in sagittal and coronal planes have been most commonly used as a standard technique of 'Department ofradiology, Kangbuk Samsung Hosp ital 2D epartm en t 이 O rth o p e di c Surgery, Kangbuk Samsung Hospital Received November 6,1995 ; Accepted March 19,1996 Address reprint requ ests to: Dong Joo Kim, M.D., Department of Radiology, Kangbuk Samsung Hospital,! 108 Pyeong-dong, Chongro-gu, Seo ul, Korea. Tel Fax MR knee imaging, and a three- dimensional(30) FT gradient- echo sequence has also been proposed for that purpose (1-6). Advantages of the 30 over the 20 technique include an improved signal -to - noise ratio and spatial resolution for a given slice thickness ; truly contiguous and thi n slices ; and the abil ity to obtai n isotropic voxels, enabl ing data reconstruction in any plane (1, 2, 5). Prolonged scanning time, the major disadvantage of the 30 technique can in practice be reduced to an acceptable level by using it in conjunction with the gradient- echo technique. The purpose of this study was to determine whether 30FT gradient- echo imaging is accu rate in evaluating the internal structures of the knee and to compare the results ofthis with those ofthe 20FT SE technique

2 Journal of the Korean Radiological Society 1996 ; 34( 4) MATERIALS and METHODS MR images were obtained of 63 knees in 61 patients referred for suspected internal derangement of the knee. These 61 patients consisted of 45 men and 16 women, aged between 13 and 62 years (mean : 35). The two authors prospectively interpreted these MR images without knowledge of the surgical results. AII the patients had undergone arthroscopic or open surgery and the results used as gold standards in this report were as shown in Table 1. With regard to tears of the medial or lateral collateralligaments which are not accessible to evaluation by arthroscopy and usually need no operative management, we simply compared Table 1. Oiagnoses by Arthroscopy or Open Surgery Tear 01 internal structure Medial meniscus Lateral meniscus A꺼미r 떼뾰띠)Dl P) M h 1 뼈빼때따Others 27 Osteochondral fracture 19 Chondromalacia patella nu R) 5 2 the MR findings on the 20 and 3D images. MR imaging was performed with a 1.5T MR imager (Signa ; GE Medical System, Milwaukee, USA). A dedicated transmit-receive extremity coil was used in all cases. The patients were positioned supine on the magnet with their knees extended naturally. After obtaining axial T1 - weighted SE images (TRITE=500/1 0) for orientation, sagittal T1 - (TR/TE=600/15) and T2 weighted (TR/TE=2500/15, 60) SE sequences and coronal T2 - weighted (TR/TE=2500/20, 60) SE sequences with fat suppression were performed. For the fat suppressed image, we used chemical shifting selective imaging technique. The other scan parameters for these SE images were as follows : 256 X 192 matrix, one excitation, 14-16cm field ofview, and 3mm slice thickness with interslice gap of 1 mm. Gradient refocused aquisition in steady state (GRASS) is the version of the limited - flip - angle, gradient- refocused pulse sequence available on our imager. By using a 30FTtechnique, 60 contiguous axial GRASS images (TRITE/flip angle=50/14/20 0 ) were obtained with a section thickness of 0.7 mm or 1 mm, a matrix size of 256 X 192, a field of view of 14 to 16 cm, and one excitation. The actual scanning time was 7 Table 2. 30FT GRASS * vs 20FT SE **: Medial Meniscus n=63(no. of surgical results) MR sequence 30FTGRASS 20FTSE Negative (36) true( - )lfalse( +) Positive (27) true( + )/false( -) 34 / 2 26 / 1 34 / 2 23 / 4 * 3-dimensional Fourier transform gradient refocused acquisition in steady state **2-dimensional Fourier transform spin-echo Sensitivity Specificity Table 3. 30FT GRASS vs 20FT SE: Lateral Meniscus n=63(no. of surgical results) MR sequence Negative (44) true( - )/false( +) Positive (19) true( + ) / false( -) Sensitivity Specificity 30FTGRASS 20FTSE 43 / 1 18 / 1 42 / 2 17 / Fig. 1. a. A peripheral tear of the medial meniscus is suspected on PO-weighted SE image revealing an uncoverage of the posterior tibial plateau and increased signal intensity between the meniscus and the capsule b. 30 axial image demonstrates the mar ginal irregularity with increased signal in tensity (arrow) along the body-posterior horn junction of medial meniscus, which was also demonstrated on several consecutive upper and lower axial images (not shown ), strongly suggesting peripheral tear which was confirmed arthroscopically a b

3 Dong Joo Kim, et al: MR Imaging 01 the Knee m i ns. 26 secs. with the 0.7 mm - thick section and 10 mins. 12 secs. with the 1mm-thick section. After reviewing the original axial images, reformatted images in arbitrary planes different from case to case were also evaluated at the workstation, and this took between 3 and 10 minutes. MR diagnosis of a meniscal tear was based on the criteria published by Reicher et al (7) and Stoller et al (8). The meniscus was considered torn if a bright line extended to the articular surface or inner or outer edge, or if the meniscus was deformed. The ligament was diagnosed as torn based only on the primary sign of the ligament itself such as a discontinuity with gap, nonvisualization of the normal ligament, or significant sweliing with highly increased signal intensity. 80th SE and 30FT GRASS images were separately reviewed and the results were compared with surgical findings. RESULTS Of the 126 menisci, 46 (27 medial and 19 lateral) had tears and 80 were normal in surgery. Forty (23 medial and 17 lateral) meniscal tears and 76 normal menisci were correctly diagnosed on 20 SE images. Six results were false negative (tour medial and two lateral meniscil, and four were false positive (two medial and two lateral meniscil. Calculated sensitivity and specificity are presented in Tables 2 and 3. 30FT GRASS imaging correctly demonstrated 44 (26 medial and 181ateral menisci) of the 46 tears (Fig. 1, 2) and 77 (34 medial and 431ateral) of 80 normal menisci. Three results (two medial and one lateral meniscil were false negative and two (one medial and one lateral meniscus) were false positive. The sensitivity and specificity ofthe 30FT GRASS technique are presented in Table 2 and 3. Oiagnosis of the small radial tear could be facilitated by the original thin - section axial images of the 30FT GRASS technique. Meticulous reexamination of cases which resulted false negative by both techniques failed to reveal the tears shown by arthroscopy to be small tears extending to the inferior articular surfaces. In some cases, 3D axial images were better than 20 sagittal or coronal images in evaluating the extent of a tear and the relationship between the torn menisci and the fragments (Fig. 1, 2) For the diagnosis of a tear of the anterior cruciate ligament (ACL), 30FT GRASS technique was superior to 20 SE in three cases (Fig. 3, 4). 20 SE images gave Fig. 2. a, b. The sagittal2d SE images show the medial meniscal tear with inward displacement 01 the torn Iragment into the intercondylar lossa suggesting a buckethandle tear c. The 3D axial image reveals delinitely the handle 01 the bucket (arrows) displaced int the intercondylar lossa. d. On the obl 띠 ue coronal reformatted image, the inward migrated meniscal Iragment (white arrow) looks pushing the PCL (black arrow) upward a b c d

4 Journal of the Korean Radiological Society 1996; 34(4) : rise to two false positive and two false negative results, and 30 GRASS resulted in one false positive and no false negative results (Table 4). There was no difference in accuracy of diagnosis of tears of the posterior cruciate I igament, and using either technique, diagnostic sensitivity and specificity were 80% and 100%, respectively. Nor were we able to find any difference between the two techniques in the results of evaluation of medial and lateral collateral ligaments (Fig. 5). Axial images of the patellofemoral joint offered by the 30 technique facilitated the detection of chondromalacia patella. The ability to reformat the original voxel images into well - resolved images in any plane disclosed in one case a small osteochondral fracture and displaced fragment which routine 20 SE images failed to depict clearly (Fig. 6) DISCUSSION MR imaging has become a useful tool for imaging the knee joint, and several studies have proved MR to be highly accurate in delineating meniscal and ligamentous tears (9-15). These studies have used the 20 technique mostly with sagittal and coronal SE sequences, but the conventional MR technique has certain limitations. Small tears on the free edge of the meniscus, tears perpendicular to the meniscus, or tears of the nonorthogonal structure such as ACL might not be identified. The very thin, contiguous sections of volume MR imaging offer major advantages over their 20 counterparts (1, 4, 5) ; (1) while 20FT p 비 se sequences may Table 4. 30FT GRASS vs 20FT SE: Anterior Cruciate Ligament n= 63 (No. of surgical results) MR sequence 30FTGRASS 20FTSE Negative (53) true( - )/false( +) Positive (1 이 true( +)/false(-) Sensitivity Specificity 52 / / Fig. 3. The continuity of the posterior fascicle of the ACL which was confirmed surgically and by oblique sagittal reformatted image (a) is not affirmative on routine SE sagittal image (b) a Fig. 4. a. On SE T2WI, the proximal portion of the ACL looks edematous and its distal slope is decreased, suggesting ACL tear somewhere in the proximal portion b. The 3D reformatted image shows midsubstance tear (arrow) of ACL. a b

5 Dong Joo Kim, et al: MR Imaging of the Knee be bothered by cross -talk from adjacent slices, 30FT sequences excite an enti re vol ume of tissue and obtain tr 비 y contiguous slices with resultant minimal volume averaging and no lost pathologic areas in section gaps (2) The signal -to- noise ratio of 30FT GRASS image is greater than that of 20FT GRASS or SE image of comparable thickness, this allows the 30 technique to obtain thinner slices and better in - plane resolution. (3). Because of the small voxel size, image reformation in any orthogonal and nonorthogonal planes is possible without additional scanning time or repositioning ofthe patient. In addition, the excessively prolonged scanning time which is the main disadvantage of the 30 volume technique could be compensated by coupling with gradient-echo sequence. With these, image acquisition times in the order of a second or subsecond are now common, and numerous imaging sequences, including GRASS, based on gradient-echoes have been developed. In GRASS images, short TRs (ie. TR( (12) with appropriately large flip angles produce the signal Fig. 5. The MCL tear from the femoral insertion site is easily diagnosed on both SE coronal (a) and 3D reformated (b) images. The bone bruise rev ealed as reticular hyperintensities in the lateral femoral condyle is obvious only on fat-suppressed SE T2WI(a) contrast related to T2, causing joint effusion to become bright in contrast to hypointense menisci and ligaments (16). The axial 30FT GRASS technique provides images of the menisci just as they are, and these axial plane images are similar to the views seen by the arthroscopic surgeon (2). In this study, axial 30FT GRASS effectively identified meniscal tears and proved its superiority to the 20 SE technique in detecting small radial or peripheral tears. Both the axial imaging plane and the thinner slice thickness with resultant diminution ir) volume - averaging may have been responsible for this. The extent and relationship between torn menisci and fragments were often better delineated on axial than on coronal and sagittal images. In the axial plane, fragments were usually displaced and horizontal tears parallel to the imaging plane or nonperipheral tears involving the superior or inferior aspects of the menisci may be difficult to detect. We found no such case, however, and the reformatted images into sagita b a b c Fig. 6. a. 3D reformatted image shows osteochondral defect (white arrows) and anteriorly displaced fragment (black arrowhead) b, c. Retrograde reevaluation of coronal (b) and sagittal (c) SE images may be suggeting the detached fragments as an ovoid low signal intensity (arrows), however, the osteochondral defect area is skipped. m x

6 Journal of the Korean Radiological Society 1996: 34(4) : tal or coronal planes might compensate for these possible shortcomings. With a sensitivity of 1 00% and a specificity of 98.1 %, the 3D protocol was accurate in the evaluation of the ACL ; this might be because of good resolution of reconstructed images in any arbitrary planes from original thin contiguous axial slices. ACL is a complicated structure of the knee, thinner than other knee ligaments and with a nonorthogonal orientation. On SE imaging, particularly with thicker slices, slight rotational changes in positioning or the presence of an effusion may substantially affect the visibility of the ligament (14). Because it is able to produce high-resolution reformatted images in any 이 ane, the 3D volume technique offers advantages in the evaluation of a nonorthogonal structure such as ACL. With regard to other ligamentous tears and other cartilaginous or osteochondral lesions, detectabilities of both techniques were same, even though 3D reformatted images tended to disclose more clearly subtle changes such as a small osteochondral fracture with anteriorly displaced fragment (Fig. 6). Because of the limited number of cases, we are unable to reach any conclusion concerning the reliability of 30FT GRASS in diagnosing lesions of the articular cartilage. The in-plane resolution of gradient- echo sequence has been known to be insufficientto consistently evaluate low- grade (grade 0 and 1) lesions correctly, even though it can separate cartilage from adjacent tissues and clearly depict high - grade lesions (4, 17, 18). Bone marrow abnormality was not evaluated. Trabecular bone imaged by 3D GRASS was lower in signal intensity and appeared coarser than that imaged by SE technique, probably due to the effect of increased magnetic susceptibility (4); gradient- echo images were less sensitive than fat - suppressed SE images in their depiction of bone bruise (Fig. 5) In summary, a 30FT GRASS sequence revealed itself at least comparable to a conventional SE sequence in the eval uation of internal structu res of the knee joi nt. It offered su perior sensitivity and specificity i n diagnosing tears of menisci and of ACL, and took only about 7.5 or 10 minutes for scanning and an additional several minutes for image reformation; this might not be a problem with regard to patient throughput. We therefore concluded that a 30FT GRASS sequence can be expected to supplement the standard techniques of MR knee i REFERENCES 1. Araki Y, Ootani F, Tsukaguchi 1, et al. MR diagnosis 01 meniscal tears 01 the knee: value 01 axial three-dimensional Fouriertranslorm GRASS images. AJR 1992 : 158: Aubel S, Heyd RL, Thaete FL, Wozney P. MR knee imaging : axial 30FT GRASS pulse sequence versus spin-echo imaging lor detecting meniscal tears. Magn Reson Imaging 1992; 1 이 4) Reeder JO, Matz SO, Becker L, Audelman SM. MR imaging olthe knee in the sagittal pr 이 ection : comparison 01 three-dimensional gradient-echo and spin-echo sequences. AJR 1989; Tyrrell RL, Gluckert K, Pathria M, Modic MT. Fast three-dimensional MR imaging 01 the knee : comparison with arthroscopy Radio/ogy 1988 : 166 : Spritzer CE, Vogler JB, Martinez S, et al. MR imaging olthe knee preliminary results with a 30FT GRASS pulse sequences. AJR 1988 ; 150 : Haggar AM, Froelich JW, Hearshen 00, Sadasvan K. Meniscal abnormalities 01 the knee: 30FT Fast-scan GRASS MR imaging AJR1988 ; 150: Reichen MA, Hartzman S, Ouckwiler GR, Bassett LW, Anderson LJ, Gold RH. Meniscal i 미 uries: detection using MR imaging Radiology 1986; 59: Stoller OW, Martin C, Crues JV 111, Kaplan L, Mink JH. Meniscal tears: pathologic correlation with MR imaging. Radio/ogy 1987; 163 : Manaster BJ. Magnetic resonance imaging 01 the knee. Semin ultrasoundct MR 1990; 11 (4) : Kelly MA, Flock T J, Kimmel JA, et al. MR imaging 01 the knee clarilication 01 its role. Arthroscopy 1991 ; 7(1): Lee JK, Yao L, Phelps CT, Wirth CR, Crajka J, Lozman J. Anterior cruciate ligamenttears: MR imaging compared with arthroscopy and clinical tests. Radiology 1988; 166: Glashow JL, Katz R, Schneider M, Scott WN. Oouble-blinded assessm ent 01 the value 01 magnetic resonance imaging in the diagnosis 01 anterior cruciate and meniscal lesions. J Bone Joint Surg 1989 ;71 A : Bassett LW, Grover JS, Seeger LL. Magnetic re sonance imaging 1 knee trauma. Skeletal Radio/1990: 19 : Mink JH, Levy T, Crues JVII I. Tears 01 the anterior cruciate ligament and menisci 01 the knee: MR imaging evaluation. Radi logy 1988 : 167: Crues JV, Mink J, Levy TL, Lotysch M, Stoller Ow. Meniscal tears 1 the knee:accuracy 01 MR imaging. Radio/ogy 1987; Price RR. Contrast mechanisms in gradient-echo imaging and an introduction to past imaging. Radiographics 1995 : 15: Recht MP, Kramer J, Marcelis S, et al. Abnormalities 01 articular cartilage in the knee: analysis 01 available MR techniques. Radiology 1993; 187 : Speer KP, Spritzer CE, Goldner JL, Garrett WE. Magnetic resonance imaging 01 traumatic knee articular cartilage i 미 - 548

7 Dong Joo Kim, et al: MR Imaging of the Knee 대한방사선의학회지 1996; 34(4) : 슬관절의자기공명영상 :3 차원퓨리어전환 GRASS 법 1 I 강북삼성병원방사선과 2 강북삼성병원정형외과 김동주 이영욱 윤은경 노인계 진승범 김준식 2 최재열 2 목적 슬관절자기공명영상방법의하나로삼차원퓨리어전환 GRASS 법으 진단적가치를알아보고자하였다. 대상및방법 : 1.5T 자기공명영상장치를이용하여 61 명의환자 63개의슬관절에서통상적인 2차원스핀에코밤법과 3차원 GRASS 방법에의한영상을얻어그결과를슬관절경이나그외수술소견에근거하여비교하였다.2차원스핀에코밤법으로는시상면의 T1 및 T2 강초영상과관상면의지밤신호억제를병행한 T2 강조영상을얻었다.3차원 GRASS 법으로는 0.7 또늠 1 mm 두께의 607H 연속횡단면영상을얻었고, 이때얻은영상과필요에따라여러방향으로재구성한영상을함쩨관찰하였다. 결과 :3차원 GRASS 법은 2차원스핀에코법에서발견하지못한 3예의내측반월판, 1 예의외측반월판, 3예의전방십자인대파열을진단하였고, 외측반월판과전방십자인대에있어서는특이도에서도우월한결과를보였다. 그외인대의파열이나글연글성병변의진단에있어서늠의미있는차이점은발견할수없었다.3차원 GRASS 법은반월판파열의정도나모앙판독에있어서우수하였다. 결론 :3차원 GRASS 법은슬관절내부구조이상의진단에있어적어도 2차원스핀에코법에견줄만하며, 특히반월판이나인대파열유무의진단이통상적 2차원스핀에코법으로어려운경우보조적자기공명영상법으로유용하리라사료된다

8 딩1. 시상부문 : 의학부분 2. 수상후보자의업적및자격 엽적기준 -창조, 탐구, 봉사정신무로국가빛인류발전에탁월한엽적을남긴자연구업적이사회의귀감이되고해당분야및사회적으로평가를받고있는자 -공공의이익에현저히공허한자 수상후보자의자격 -한국인및한국계인사로서추천일현재생존해있는자 3. 후보자추천인자격 호암상위원회위원 본위원회에서추천인으로위촉한분 ( 단체장및전문인사 ) 국제적인권위를지난상의수상자 호암상수상자및심사위원을역임한자 4. 구비서류 추천서 ( 본위원회소정양식, 학회비치 ) 2부 엽적증빙자료 ( 대표논문 3편이내 ) 5. 추천서접수및발표 접수기한 : 199 6년 9월 16일까지 접수처 : 본학회사무국또는호암상위원회접수처 발표및시상 : 1997년 2월일간지에발표후 3월 22일시상예정 기타자세한사항은본학회로문의바랍니다. 쉽터원고모집안내 쉽터의원고는방사선의학또는영상과관계가있는사진또는그럼으로서독자들에게흥미를줄수있는내용이어야합니다. 한쪽단위가되어야하며두쪽이상인경우는독립적으로가치가있어야하며, 게재여부는대한방사선의학회편집위원회에서결정하며게재료는학회에서부담합니다. n U