| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Musculoskeletal Imaging |
1 From the Dept of Radiology, Brigham and Women’s Hospital Harvard Medical School, 75 Francis St, ASB-1, L-1, Room 003E, Boston, MA 02115 (H.Y., P.L.); Depts of Radiology (K.S.) and Immunology and Rheumatology (M.G.), Stanford Univ School of Medicine, Stanford, Calif; and Sports Orthopedic and Rehabilitation Medicine Associates, Menlo Park, Calif (M.F.D.). Received Apr 22, 2002; revision requested Jun 21; final revision received Aug 13, 2003; accepted Oct 8. Supported in part by grants from the Chiron Corporation, the Whitaker Foundation, and the Japanese Overseas Research Fellowships from Monbusho (the Ministry of Education, Science, Sports, and Culture of Japan). Address correspondence to P.L. (e-mail: pklang@partners.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Sagittal fat-suppressed intermediate-weighted fast spin-echo (FSE) (repetition time msec/echo time [TE] msec, 4,000/13), sagittal T2-weighted FSE (4,000/39), and sagittal fat-suppressed three-dimensional (3D) spoiled gradient-echo (SPGR) (60/5, 40° flip angle) MR images were acquired in 28 patients and four volunteers. FSE images with a TE of 13 msec were considered "short-TE images"; those with a TE of 39 msec were considered "long-TE images." Presence of normal MR imaging appearance of articular cartilage was determined by one author. Contrast between cartilage and adjacent structures (meniscus, joint capsule, synovial fluid, muscle) was calculated in posterior regions of the femoral condyle on images obtained with each sequence; Wilcoxon signed rank testing was performed.
RESULTS: The following appearances were observed in patients with knee osteoarthritis (on short-TE FSE, long-TE FSE, and SPGR MR images, respectively): (a) ambiguity of surface contour in posterior region of the femoral condylar cartilage (in zero, zero, and 20 patients), (b) linear area of high signal intensity in deep zone adjacent to subchondral bone of femoral condyle (in zero, zero, and 26 patients), (c) pseudolaminar appearance in posterior region of femoral condylar cartilage (in seven, nine, and 24 patients), (d) truncation artifact in patellofemoral compartment (in seven, six, and 27 patients), (e) susceptibility artifact on cartilage surface caused by air or metal (in three, three, and 11 patients), (f) decreased signal intensity in distal part of trochlear cartilage (in 28, 28, and 28 patients), (g) cartilage thinning adjacent to the anterior horn of the lateral meniscus (in 19, 19, and 21 patients), and (h) focal cartilage flattening in posterior region of femoral condyle (in 16, 16, and nine patients). Cartilage-meniscus and cartilage–synovial fluid contrast was significantly higher on fat-suppressed FSE than on fat-suppressed 3D SPGR MR images (P < .001).
CONCLUSION: Fat-suppressed FSE and 3D SPGR MR images showed nonuniform signal intensity arising from articular cartilage and cartilage thinning, both of which could mimic disease.
© RSNA, 2004
Index terms: Arthritis, 452.771 • Knee, anatomy, 452.121411, 452.121412, 452.121415 • Knee, ligaments, menisci, and cartilage, 452.121411, 452.121412, 452.121415 • Knee, MR, 452.121411, 452.121412, 452.121415 • Magnetic resonance (MR), artifact
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Magnetization transfer contrast imaging can "separate" articular cartilage from adjacent joint fluid by suppressing the signal produced by cartilage (2–4). Fast spin-echo (FSE) MR imaging with fat suppression for acquisition of intermediate- and T2-weighted images depicted articular cartilage abnormalities in osteoarthritis with good accuracy when its results were compared with those of arthroscopic grading (5,16). Fat-suppressed three-dimensional (3D) spoiled gradient-echo (SPGR) imaging has been reported to be a more sensitive imaging sequence than standard T1-weighted spin-echo and T2-weighted spin-echo sequences for the detection of articular cartilage defects in the knee (6–8).
Results of recent studies have shown that driven-equilibrium Fourier transform imaging provides contrast between cartilage and joint fluid by enhancing the signal from joint fluid, rather than by suppressing the signal from cartilage as some sequences do (9). The cartilage–synovial fluid contrast-to-noise ratio of driven-equilibrium Fourier transform is up to four times greater than that of FSE and SPGR sequences. Delayed gadolinium diethylenetriaminepentaacetic acid2-–enhanced MR imaging is also a promising method that has potential for monitoring the glycosaminoglycan content of cartilage in vivo (10,23).
The relative signal intensity of the normal articular cartilage is dependent on the pulse sequences that are used. T2-weighted spin-echo imaging, intermediate-weighted and T2-weighted FSE imaging, magnetization transfer contrast imaging, and driven-equilibrium Fourier transform imaging can depict high-signal-intensity synovial fluid and intermediate- to low-signal-intensity cartilage, while fat-suppressed 3D SPGR sequences show bright cartilage and dark synovial fluid.
However, the signal intensity of the normal articular cartilage may not be uniform owing to such artifacts and other phenomena as the magic angle effect, truncation artifacts, chemical shift artifacts, magnetic susceptibility effects, and regional anatomic variations (13,14). Although the anatomic distribution and orientation of collagen in articular cartilage enables one to delineate anatomic zones at microscopy (15), a laminated appearance of the articular cartilage at MR imaging is strongly influenced by the anisotropic arrangement of the collagen fibers and the magic angle effect (24).
Some researchers have suggested that in healthy volunteers, the laminar appearance within the articular cartilage on fat-suppressed 3D SPGR images is predominantly attributable to truncation artifact rather than to histologic zonal anatomy (25,26). Thus, the MR imaging appearance of the articular cartilage is highly variable, and understanding the normal variations is clinically important for improving diagnostic accuracy and avoiding misdiagnoses. The purpose of this study was to evaluate normal MR imaging findings that may mimic disease of the articular cartilage in healthy subjects and in patients with osteoarthritis of the knee.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This study was approved by the institutional review board of Stanford University, and each volunteer and patient gave written informed consent. The patients were participating in a phase II clinical trial evaluating the efficacy of a chondroregenerative drug, recombinant human insulin-like growth factor I, for the treatment of osteoarthritis. This protocol consisted of a phase II, randomized, double-blind, placebo-controlled, multicenter, dose- and regimen-finding study and included arthroscopic correlation.
Inclusion criteria for our study were as follows: (a) age of 40 years or older, (b) presence of a degenerative articular cartilage lesion of the knee of grade III or lower in a weight-bearing area documented at baseline arthroscopy (27) and in a location appropriate for cartilage biopsy, (c) body weight of more than 50 kg (110 lbs) and less than 110 kg (242 lbs), and (d) full or partial weight-bearing status preoperatively, with or without the use of assistive devices such as walkers, crutches, or canes.
Medical exclusion criteria were as follows: (a) clinically important uncontrolled cardiac, respiratory, gastrointestinal, renal, neurologic, hepatic, or psychiatric illness at the time of enrollment or within the preceding 6 months; (b) any suspicion of important immunocompromise (eg, human immunodeficiency virus infection, Cushing disease, neutropenia, severe alcohol or drug abuse); (c) any systemic immunosuppression or chemotherapy within the preceding 6 months (180 days); (d) a diagnosis of type 1 or type 2 diabetes mellitus; (e) any history of seizures or syncope within the preceding 6 months (180 days); (f) any history of malignancy within 10 years, with the exception of cured basal cell carcinoma; (g) any suspicion of malignancy after clinical history and physical examination, stool guaiac test, chest radiography, and urinalysis; (h) a prostate-specific antigen level (men only) of 4 ng/mL or greater within 2 weeks of the study; (i) a history of marked allergic reaction to yeast or yeast-derived products; (j) a recent history or presence of thrombocytopenia or hemorrhagic diathesis (platelet count <20,000); or (k) therapy with heparin or warfarin for systemic anticoagulation.
Orthopedic exclusion criteria were as follows: (a) arthroscopy or open knee surgery within the previous 3 months (90 days); (b) open knee surgery at the time of index arthroscopy for enrollment in the clinical trial; (c) torn or ruptured anterior cruciate ligament; (d) axial malalignment, whether untreated or treated at the time of the baseline arthroscopy (no more than 0° of varus or 15° of valgus as measured on a conventional anteroposterior radiograph of the knee); (e) ligamentous instability treated at the time of baseline arthroscopy; (f) chronic knee effusion requiring repeated arthrocentesis (ie, more frequently than once every 3 months); (g) evidence of knee joint infection at clinical examination; (h) any documented history of inflammatory arthritides or substantial crystalline arthritis or chondrocalcinosis at arthroscopy; (i) intraarticular injection of hyaluronic acid within the previous 3 months (90 days); or (j) intraarticular injection of glucocorticoids within the previous month (30 days).
Subjects with any contradindication to MR imaging identified by the investigator or defined by the institutional guidelines were also excluded from the study. The clinical trial included a baseline MR imaging examination and follow-up MR imaging after 13 weeks of treatment with either placebo or recombinant human insulin-like growth factor I administered at two different doses. However, results from only the baseline MR imaging examination performed before administration of the placebo or the chondroregenerative drug were included in this study. Arthroscopy was performed 2–26 days (mean, 6.29 days) before baseline MR imaging in 17 patients participating in the trial.
MR Imaging
All MR images were obtained with a 1.5-T MR imaging unit (Signa, software version 8.2.5; GE Medical Systems, Milwaukee, Wis) by using a standard transmit-receive knee coil provided by the manufacturer. In each volunteer and patient, sagittal fat-suppressed intermediate-weighted FSE (repetition time msec/echo time [TE] msec, 4,000/13), sagittal fat-suppressed T2-weighted FSE (4,000/39), and sagittal fat-suppressed 3D SPGR (60/5, 40° flip angle) sequences were performed. We called the FSE sequence with a TE of 39 msec a "long-TE FSE sequence" and the FSE sequence with a TE of 13 msec a "short-TE FSE sequence" for the sake of simplicity, although the TE of 39 msec was relatively short compared with the TE of a conventional long-TE sequence.
Both fat-suppressed FSE sequences were obtained with an echo train length of eight, a 4-mm section thickness, a 1-mm intersection gap, an image matrix of 256 x 192, a 120-mm field of view, and two signals acquired, for an acquisition time of 3 minutes 12 seconds per sequence. Contiguous fat-suppressed 3D SPGR images were obtained with a 1.5-mm section thickness, an image matrix of 256 x 160, a 120-mm field of view, and one signal acquired, for a total acquisition time of 10 minutes 18 seconds.
Image Evaluation
A previous study (28) revealed a number of findings that could potentially mimic disease of the articular cartilage. Our aim in this study was to focus on those findings and determine their incidence. Our study, therefore, involved evaluation of the presence of each of the following: (a) ambiguity of the surface contour in the posterior region of the femoral condylar cartilage (meaning that the surface of the articular cartilage is not clearly defined and is indistinguishable from surrounding soft-tissue structures); (b) a linear area of high signal intensity in the deep zone adjacent to the subchondral bone of the femoral condyle; (c) a pseudolaminar structure in the posterior region of the femoral condylar cartilage; (d) truncation artifact in the patellofemoral compartment; (e) susceptibility artifact on the cartilage surface caused by air or metal; (f) decreased signal intensity in the distal part of the trochlear cartilage; (g) cartilage thinning in the central portion of the lateral femoral condyle (the overlying concavity of the femoral contour adjacent to the anterior horn of the lateral meniscus); and (h) focal cartilage flattening in the posterior region of the femoral condyle.
The above findings were evaluated in volunteers and patients by one author (H.Y.). Twenty-four of the 28 patients had a history of knee arthroscopy or surgery. Seventeen of the 28 patients had undergone arthroscopy prior to MR imaging, and the qualitative analyses were performed only for areas of normal cartilage seen at arthroscopy. We compared arthroscopic and MR imaging findings of normal cartilage and abnormalities in these patients. The findings in the remaining patients who had not undergone arthroscopy were documented solely at MR imaging.
Quantitative analyses were performed in 28 patients to evaluate the ambiguity of the surface contour in the posterior region of the femoral condylar cartilage. One image of the posterior regions of the medial and lateral femoral condyles was selected for each patient by one author (H.Y.). Of the 56 regions selected, only 45 could be measured. In the remaining regions, the visualized areas of cartilage or meniscus were not large enough to enable calculation of signal intensity reliably due to full-thickness cartilage loss, cartilage thinning of greater than 50%, or partial or complete meniscectomy.
For the 45 regions, we visually selected one section on which ambiguity of the surface was shown or on which the surface showed signal intensity that was similar to that of the adjacent structure. Then, the signal intensities of the cartilage, the meniscus, and the soft-tissue structures (ie, joint capsule, synovial fluid, or muscle) adjacent to the cartilage were measured. Corresponding sections containing these tissues were selected from among the fat-suppressed intermediate-weighted and T2-weighted FSE images and the fat-suppressed SPGR images. One author (H.Y.) calculated signal intensities by placing a cursor containing 20–30 pixels over a region of interest. The regions of interest for each subject were placed in identical positions on corresponding sections.
The contrast between the cartilage and the meniscus or between the cartilage and the adjacent structures was determined for each sequence with one of the two following formulas: (SIc - SIm)/(SIc + SIm) or (SIc - SIa)/(SIc + SIa), where SIc is the signal intensity of the cartilage, SIm is the signal intensity of the meniscus, and SIa is the signal intensity of an adjacent structure. The possible values of the contrast ranged from 1.0 to -1.0. Positive cartilage-to-meniscus or cartilage-to–adjacent structure contrast values indicated that the cartilage was brighter than other tissues, while a value of zero indicated that there was no signal intensity difference between the two tissues.
Statistical Evaluation
Contrast was reported in terms of mean values and SDs. The Wilcoxon signed rank test was used to determine the significance of differences in contrast between fat-suppressed FSE images and fat-suppressed 3D SPGR images. For each analysis, a P value of less than or equal to .05 was considered to indicate a significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Numerous MR imaging pulse sequences have been used for the diagnosis of cartilage disease. The use of fat-suppressed 3D SPGR MR imaging for the detection of cartilage lesions has been shown to result in a sensitivity of 75%–85% and a specificity of 97% (6). Recht et al (17) reported that fat-suppressed 3D SPGR had a sensitivity of 96%, a specificity of 95%, and an accuracy of 95% in cadaveric knees at 1.5-T MR imaging. Recht et al (8) also reported that with fat-suppressed 3D fast low-angle shot MR imaging, the sensitivity was 81%, the specificity was 97%, and the accuracy was 97% for detection of abnormal cartilage in the patellofemoral joint at 1.0 T. The sensitivity and specificity of FSE imaging were observed to be 87% and 94%, respectively, in a study involving arthroscopic correlation (16). Another study revealed that sensitivity and specificity with the combination of transverse and coronal T2-weighted FSE sequences with fat suppression were 93% and 99%, respectively (21).
In this article, we report several normal MR imaging appearances of articular cartilage that were seen on both fat-suppressed FSE images and fat-suppressed 3D SPGR images. The signal intensity of articular cartilage was not uniform, particularly on fat-suppressed 3D SPGR images. Some well-recognized normal variants that mimic disease—in addition to the ones previously reported by Waldschmidt et al (13,14) and Disler et al (19)—were observed in this study. Ambiguity of the surface contour of the cartilage surface in the posterior region of the femoral condyle was invariably seen on fat-suppressed 3D SPGR images. According to the results of quantitative analysis, this finding appears to be due to low contrast between cartilage and meniscus or adjacent structures such as the joint capsule, synovial fluid, and muscle.
On fat-suppressed FSE images, the mean contrast between cartilage and the joint capsule or muscle approached zero, like the mean contrast on fat-suppressed 3D SPGR images. Nonetheless, the cartilage contour was qualitatively more clearly seen on fat-suppressed FSE images than on fat-suppressed 3D SPGR images (Figs 1, 2). The contrast between cartilage and meniscus or adjacent structures at FSE imaging was produced as follows: (a) When the meniscus or a normal thin synovial capsule was adjacent to cartilage, the contrast had positive values; (b) when synovial fluid was apposed to cartilage, the contrast had high negative values; and (c) even when muscle or a thickened or inflamed capsule that showed signal intensity similar to that of cartilage was next to cartilage, a linear area of low signal intensity existed between cartilage and these structures (Figs 1, 2).
Therefore, although the mean contrast between cartilage and the joint capsule or muscle on FSE images approached zero numerically, differentiation of the cartilage from adjacent structures on FSE images was still possible. In the same posterior region on SPGR images, a linear area of high signal intensity adjacent to the subchondral bone was also frequently seen; this phenomenon tended to result in a complex appearance of the signal intensity from cartilage in this region. The origin of this linear area of high signal intensity is uncertain. However, if there is a cartilage defect in this area, an accurate diagnosis may be difficult. Hence, it is important to pay close attention when assessing cartilage abnormalities in the posterior region of the femoral condyle with fat-suppressed 3D SPGR imaging. The finding described above also represents a major pitfall in the segmentation of cartilage for measurement of global cartilage volume (33) and for generation of 3D thickness maps (34).
In the present study, truncation artifact was also frequently observed in both the patellofemoral compartment and the posterior region of the femoral condyles on fat-suppressed 3D SPGR images, although previous researchers reported observing this artifact only in the region of the patellofemoral joint (13,25,26). In addition, because other structures such as the synovial capsule, tendon, and muscle were closely apposed to the articular cartilage in the posterior region of the femoral condyles, multiple layers of low signal intensity were sometimes recognized on SPGR images in our study. It should be noted that this multilaminar structure does not indicate degenerative change of the articular cartilage, nor does it reflect the anatomic layers of the cartilage. Interestingly, FSE images sometimes showed a truncation artifact in the patellofemoral compartment that appeared as a linear area of high signal intensity in the cartilage.
Decreased signal intensity in the distal part of the trochlear cartilage was seen on images obtained with all sequences in all volunteers and patients with osteoarthritis. High signal intensity in cartilage proximal to this region resulted from the magic angle effect. Although this finding was identified as a normal signal pattern in a previous report (5), there were few comments regarding its incidence and cause. The cause of the very low signal intensity in this region is not clear, but presumably is related to the anisotropic arrangement of collagen fibers.
A decreased cartilage thickness, in comparison with the cartilage thickness in the more anterior or posterior regions of the lateral femoral condyle, was also frequently observed above the anterior horn of the lateral meniscus. In a study by Disler et al (19), all 43 patients referred for complaints of knee pain had smoothly contoured thinning of cartilage in the same region (the lateral femoral notch), and Disler et al conluded that this finding was normal. This finding was seen predominantly in the lateral condyle. The articular cartilage of the medial femoral condyle is not situated immediately adjacent to the anterior horn of the medial meniscus but rather posterosuperiorly to the meniscus. Thus, this appearance of thinned cartilage in the lateral femoral condyle could be related to weight bearing. It is important to recognize this normal variant to distinguish it from true thinning resulting from degenerative change or trauma in this region. Smooth transition of the cartilage thickness is one of the characteristics that differentiate healthy cartilage from true cartilage defects. The signal intensity in this area sometimes appears dark, probably due to the anisotropic arrangement of collagen fibers. However, in our study, high signal intensity was never seen in this region unless there was underlying cartilage disease.
Cartilage thinning observed in the posterior region of the femoral condyle may be caused by a mechanism other than weight bearing. In our study, this focal cartilage flattening was observed in both the medial and lateral femoral condyles. The structures adjacent to the medial condyle consist of the muscle belly and tendon of the medial head of the gastrocnemius and the posteromedial joint capsule. The structures adjacent to the lateral condyle consist of the muscle belly and tendon of the lateral head of the gastrocnemius, the muscle belly and tendon of the plantaris muscle, and the posterolateral joint capsule. The posterior regions of the femoral condyles are closely apposed to these structures, but the role they have in focal cartilage thinning is uncertain. This relationship may be relevant to the ambiguous surface contour observed in the posterior regions of the femoral condylar cartilage on SPGR MR images in our study.
Our study had several limitations. All images were obtained in the sagittal plane. However, sagittal images are superior to coronal or transverse images for assessment of the knee joint because sagittal sections can show all of the cartilage components of the knee and are the most frequently used sections for evaluating internal derangement in clinical settings. Truncation artifact may have been diminished or decreased if we had increased the image matrix size, but this would have been at the expense of temporal resolution. It may have been helpful to include more healthy volunteers to demonstrate these artifacts. Since most of the subjects imaged were patients who already had osteoarthritis, some of the artifacts seen could have been real abnormalities. However, the use of healthy subjects for evaluating cartilage-sensitive pulse sequences has its own limitations: As the cartilage degenerates, its signal may change, thereby influencing the appearance of the cartilage and any related artifacts. Thus, it is generally preferable to perform any analysis of signal intensity, image contrast, and image artifacts in patients with osteoarthritis—that is, the clinical target group—rather than in healthy volunteers.
Seventeen of the patients in our series had undergone invasive arthroscopy prior to the MR imaging examinations. Arthroscopy has the potential to affect the appearance of cartilage at MR imaging. Also, the number of patients for whom results of arthroscopic correlation were available was small. Finally, we investigated MR imaging appearances by using only fat-suppressed FSE imaging and fat-suppressed 3D SPGR imaging. It would also be useful to consider whether other techniques such as driven-equilibrium Fourier transform and T1-weighted imaging with intravenous gadolinium chelate administration would depict similar features or demonstrate other "normal" appearances. Despite these limitations to our study, we believe it is important to recognize these normal imaging variants because the MR imaging appearances demonstrated in this study were common in both healthy volunteers and in patients with osteoarthritis and sometimes obscured coexistent cartilage degeneration.
In summary, fat-suppressed FSE MR images and fat-suppressed 3D SPGR MR images showed nonuniform signal intensity arising from the articular cartilage and a high frequency of normal variations that mimicked disease, including (a) ambiguity of the surface contour in the posterior region of the femoral condylar cartilage, (b) a linear area of high signal intensity in the deep zone adjacent to the subchondral bone of the femoral condyle, (c) a pseudolaminar appearance in the posterior region of the femoral condylar cartilage, (d) truncation artifact in the patellofemoral compartment, (e) susceptibility artifact on the cartilage surface caused by air or metallic fragments, (f) decreased signal intensity in the distal part of the trochlear cartilage, (g) cartilage thinning adjacent to the anterior horn of the lateral meniscus, and (h) focal cartilage flattening in the posterior region of the femoral condyle. Features a through e were predominantly observed on SPGR images, and the latter three were seen equally often on images obtained with all sequences. It is clinically important to recognize these imaging features to achieve an accurate diagnosis of articular cartilage disease.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, H.Y., P.L.; study concepts and design, H.Y., P.L.; literature research, H.Y., P.L.; clinical studies, H.Y., K.S., M.G., M.F.D.; data acquisition, H.Y., K.S., M.G., M.F.D.; data analysis/interpretation, H.Y., P.L.; statistical analysis, H.Y., P.L.; manuscript preparation, editing, and revision/review, H.Y., K.S., P.L.; manuscript definition of intellectual content and final version approval, H.Y., P.L.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. S. Bauer, S. J. Krause, C. J. Ross, R. Krug, J. Carballido-Gamio, E. Ozhinsky, S. Majumdar, and T. M. Link Volumetric Cartilage Measurements of Porcine Knee at 1.5-T and 3.0-T MR Imaging: Evaluation of Precision and Accuracy. Radiology, November 1, 2006; 241(2): 399 - 406. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. P. Recht, D. W. Goodwin, C. S. Winalski, and L. M. White MRI of Articular Cartilage: Revisiting Current Status and Future Directions Am. J. Roentgenol., October 1, 2005; 185(4): 899 - 914. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
