| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Cardiac Imaging |
1 From the Departments of Radiology (F.S., F.Z.) and Cardiology (G.M., M.B.), University of Genoa School of Medicine, Largo R Benzi 10, Genoa I-16132, Italy. From the 1997 RSNA scientific assembly. Received September 9, 1998; revision requested October 26; final revision received June 25, 1999; accepted July 22. Address reprint requests to F.S., Department of Diagnostic Imaging, the Biomedical Institute, Via Prà 1/b, I-16157, Genoa, Italy (e-mail: sardanelli@biomedical.www-it.com).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Forty-two patients (age range, 50–79 years) underwent MR coronary angiography (1.5 T). A navigator-echo sequence was used. Two or three 15% overlapped transverse slabs were acquired. Data were analyzed by readers blinded to conventional coronary angiographic results. On conventional coronary angiograms, coronary arterial stenoses of 50% or greater narrowing were considered significant. On MR coronary angiograms, the major coronary vessels were subdivided into proximal (within 5 cm) and distal (beyond 5 cm) segments, except for the left main vessel. Stenoses of 50% or greater were identified on reformatted multiplanar MR coronary angiograms.
RESULTS: Three MR coronary angiographic examinations were aborted because of patient claustrophobia; 39 of 39 left main, 117 of 117 proximal, and 78 of 117 distal segments were visualized. MR coronary angiography showed a sensitivity of 82% (95% CI: 73%, 91%) and a specificity of 89% (95% CI: 85%, 94%) in overall stenoses identification, of 90% (95% CI: 81%, 99%) and 90% (95% CI: 83%, 96%) for proximal segments, and of 68% (95% CI: 50%, 86%) and 81% (95% CI: 71%, 92%) for distal segments, respectively.
CONCLUSION: Navigator-echo, 3D MR coronary angiography is a promising sequence for assessing coronary arterial stenoses, but further improvements are required for distal segments.
Index terms: Coronary angiography, comparative studies, 54.12142, 54.1244 • Coronary vessels, diseases, 54.762 • Coronary vessels, MR, 54.121412, 54.121415, 54.12142 • Coronary vessels, stenosis or obstruction, 54.762 • Magnetic resonance (MR), three-dimensional, 54.12142 • Magnetic resonance (MR), vascular studies, 54.12142
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
On the basis of these factors, investigators are looking for a noninvasive diagnostic alternative to conventional coronary angiography. Magnetic resonance (MR) imaging has been used since 1987 for imaging coronary arteries (14). Different pulse sequences have been proposed to optimize MR coronary angiography. In particular, two-dimensional, breath-hold, fat-saturated, segmented–k-space, gradient-echo techniques have been used by different authors (2,3,15–25).
However, with the two-dimensional approach, several breath holds are necessary to encompass the entire coronary arterial tree, since a different section orientation is required for each vessel; moreover, this approach permits only a low spatial resolution, with poor depiction of small and tortuous vessels and with the potential for incorrect evaluation of stenoses (2,3,11,13,18,21,22,24–29).
Recently, a three-dimensional (3D), electrocardiographically triggered and respiratory-gated, fat-saturated gradient-echo sequence (3D navigator echo) has been developed and applied to evaluate native coronary arteries (12,29–34) and coronary arterial bypass grafts (34). This sequence is less operator and patient dependent and requires less imaging time than two-dimensional MR coronary angiography (31,34).
The purpose of our study was to assess the value of 3D, navigator-echo MR coronary angiography in assessing stenoses of the major epicardial coronary arteries, with conventional coronary angiography as the standard of reference.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Exclusion criteria for MR coronary angiography were coronary intervention (balloon angioplasty or stent placement) during coronary catheterization, atrial fibrillation, frequent ventricular ectopic rhythms, hemodynamic instability, or general contraindications to MR imaging. The study was approved by the institutional review board. All patients gave written informed consent.
Conventional Coronary Angiography
Conventional coronary angiography was performed with a CGR Arcomax FMA unit (GE Medical Systems, Milwaukee, Wis); a femoral arterial approach was used for all patients. Iodinated contrast agent (iomeprol; [Iomeron 300; Bracco, Milan, Italy]) was injected after selective catheterization of the right and the left coronary arteries. Left and right anterior oblique images were obtained for both left and right coronary arteries. A total of 50–100 mL of iodinated contrast agent (15–30 g of iodine) was used for each patient. Conventional coronary angiograms were analyzed by consensus by two independent readers (F.Z., M.B.) who were not involved in the MR imaging procedure. Stenoses of 50% or greater narrowing were considered hemodynamically significant. The presence, site, and degree of stenosis were recorded.
MR Coronary Angiography
MR coronary angiography was performed with a superconducting 1.5-T unit (Vision; Siemens Medical Systems, Erlangen, Germany) with a 25 mT/m gradient and a 0.6-msec rise time. The built-in body coil was used, with the patient in a supine position. Leads for electrocardiographic triggering were placed on the patient's back.
After a multiplanar localizing scout image was obtained, two saturation slabs parallel to the z axis (sagittal and oblique) were positioned on the right side of the diaphragmatic dome to monitor the position of the diaphragm during the respiratory cycle. Then a navigator scout image was acquired in about 20 seconds to estimate the mean time of the respiratory cycle through two-dimensional imaging of the diaphragmatic position. In fact, the number of signals acquired in the subsequent 3D, navigator-echo sequence had to be selected to make sure that enough measurements were used to have at least one of them accepted per respiratory cycle (the number of signals acquired multiplied by the mean R-R distance had to be larger than the respiratory cycle). In practice, five signals acquired for the 3D, navigator-echo sequence fulfilled this requirement in most patients; four or six signals were acquired in only a few patients.
To optimize spectral fat saturation, a shimming procedure was performed until a field homogeneity with a tolerance of 0.4 ppm was obtained. On this basis, a fat-selective saturation pulse was applied during every cardiac cycle to cancel the MR imaging signal from the epicardial fat that surrounds the coronary arteries.
MR coronary angiography was performed with a 3D, fat-saturated, gradient-echo sequence that was non–velocity-compensated, was electrocardiographically triggered, and was retrospectively respiratory-gated (3D navigator echo). Technical parameters were a repetition time of 7.4 msec and an echo time of 2.7 msec (7.4/2.7); matrix, 128 x 256; field of view, 30 cm (square or rectangular, from 6 x 8 to 8 x 8); in-plane resolution, 1.2 x 2.3 mm; four to six signals acquired (see earlier explanation), each starting during middiastole; and 24 phase-encoding steps for each cardiac cycle with use of an incremental flip angle of 10°–90°, as already described by Li et al (12). In other words, we measured each phase-encoding step four to six times, with each acquisition saved with the navigator position.
Retrospectively, the phase-encoding measurements within the main diaphragmatic position plus or minus 2 pixels were selected. Thus, it was possible that more than one measurement was accepted for a given phase-encoding step as long as all of these were within the acceptance range. The acquisition window in middiastole was 178 msec. To image the whole coronary arterial tree, two or three 15%-overlapping, transverse, 3D slabs (slab thickness, 48 mm; 24 partitions; 2-mm effective thickness) were acquired. The total examination time was about 30 minutes.
Data Analysis
The 3D data set from the coronary arterial tree was analyzed by consensus by two investigators (F.S., G.M.) who were blinded to clinical data and conventional coronary angiographic results. Multiplanar reconstruction software on a satellite workstation (MagicView 1000; Siemens) was used. The origin of each of the four major epicardial vessels—the left main coronary artery, the left anterior descending coronary branch, the left circumflex coronary branch, and the right coronary artery—was identified by scrolling forward and backward to view subsequent reformatted sections with straight planes, and the vessels' course was followed through the 3D volume in an oscillating cine format. Normal coronary vessels were identified as high-signal-intensity linear (on longitudinal views) or roundish structures (on transverse views) within the suppressed epicardial fat.
Except for the left main vessel, coronary arteries were subdivided approximately into two segments: proximal, within 5.0 cm from their origin; and distal, beyond 5.0 cm from their origin. The conventional coronary angiographic criteria (the origin of the collateral vessels) were not used because of the uncertainty of recognition of the collateral vessels on MR coronary angiograms.
On MR coronary angiograms, a 50% or greater reduction of the luminal diameter or a signal void with distal evidence of high-signal-intensity flow was considered a severe stenosis; a signal void without evidence of distal flow was considered an obstruction. When recognized, stenoses were imaged in a two-dimensional, oblique, operator-defined plane.
The specificity, sensitivity, predictive values, and accuracy were calculated for each vessel (by considering proximal and distal segments, except for the left main coronary artery), for all of the proximal segments, for all of the distal segments, and for the entire MR coronary angiographic examination by using conventional coronary angiographic data as the standard of reference.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
At a subjective evaluation, excellent MR angiograms, with complete demonstration of the whole (proximal and distal) coronary artery tree were obtained in 24 of 39 patients. Despite the presence of moderate artifacts, a diagnostic evaluation of the proximal branches was possible in the remaining 15 patients. Postprocessing required about 45 minutes for each patient and ranged from 25 to 70 minutes.
On MR coronary angiograms, the great cardiac vein was identified in 39 of 39 (100%) patients and was always differentiated from the proximal segments of the left anterior descending and left circumflex coronary arteries by its typical diverging course. The left main coronary artery was clearly imaged in 39 of 39 (100%) patients; 117 of 117 (100%) proximal and 78 of 117 (66.7%) distal segments of the three major coronary arteries (the left circumflex, left anterior descending, and right coronary arteries) were visualized at MR coronary angiography. Detailed information regarding MR and conventional angiograms of the coronary vessels is displayed in Table 1.
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
In the past decade, an enormously wide spread of MR imaging techniques occurred, with continuous technologic development. The assessment of MR imaging as a noninvasive modality for the diagnosis of ischemic heart disease is today one of the most challenging topics in medical research. Indirect examinations such as spin-echo, gradient-echo, cine, and perfusion MR imaging of myocardium have been well assessed but still have not entered routine clinical practice (35). Until now, direct evaluation of coronary arterial disease with MR coronary angiography was still in an experimental stage (27,36). We now show that MR coronary angiography is well poised to soon become a clinically useful diagnostic test.
Important problems remain for distal segment identification with MR coronary angiography, in particular for the right coronary artery and the left circumflex coronary artery (22 of 39 segments [56%] and 25 of 39 segments [64%] visualized, respectively). Our poor results for the distal segment of the right coronary artery can be explained by considering that in those patients with dominance of the left circumflex coronary arterial segment (six patients in our study), the distal segment of the right coronary artery was very small. If we had excluded these six patients, detection in the distal segment of the right coronary artery would have increased to 22 of 33 arteries (67%).
Our difficult identification of the distal segment of the left circumflex coronary artery was probably because of the use of only straight planes for multiplanar reconstruction; curved planes would have permitted a more precise identification of such a small and tortuous vessel.
For stenosis identification, our overall values of sensitivity (82% [55 of 67]) and specificity (89% [149 of 167]) were slightly worse than the best results obtained with two-dimensional breath-hold acquisitions reported by Manning et al (18) for only proximal and middle portions of the coronary arteries (90% [44 of 49] sensitivity and 92% [90 of 98] specificity).
Four stenoses in proximal segments (three in the left circumflex and one in the right coronary artery) and eight stenoses in distal segments (three in the left anterior descending, three in the left circumflex, and two in the right coronary artery) were not identified at MR coronary angiography because of eccentric atherosclerotic plaques; their detection also was difficult at conventional coronary angiography and required multiple projections.
MR coronary angiography showed stenoses of 50% or greater in 18 segments (eight proximal and 10 distal) that proved false at conventional coronary angiography. Stenoses were not present at all in 11 segments and were less than 50% in seven segments (two in the proximal segment of the left anterior descending coronary artery and five in the distal segment of the left circumflex coronary artery). Moreover, conventional coronary angiography depicted one stenosis in the first diagonal branch and another stenosis in the obtuse marginal branch that were not detected with MR coronary angiography.
While we obtained interesting values of sensitivity and specificity, if compared with those obtained in previous experiences (2,3,18,29), for all proximal segments—and especially for the proximal segments of the left circumflex artery (sensitivity of 73% [eight of 11 segments] and specificity of 89% [25 of 28 segments])—we still find important difficulties in detecting stenoses of distal and collateral vessels, especially for the distal segment of the left circumflex coronary artery (sensitivity of 57% [four of seven segments] and specificity of 72% [13 of 18 segments]) and in the left anterior descending coronary artery (sensitivity of 67% [six of nine segments] and specificity of 86% [19 of 22 segments]). However, the different clinical effect of distal-segment coronary arterial lesions versus that of proximal-segment lesions should be remembered: The smaller the vessel, the more difficult the therapeutic approach with angioplasty, stent placement, or bypass graft surgical implantation.
Our study design with MR coronary angiography performed after conventional coronary angiography resulted in the exclusion of six patients who underwent coronary intervention during coronary catheterization and who had at least one high-grade stenosis that would have been detected easily with MR coronary angiography. Moreover, three of the other five patients excluded for different reasons (atrial fibrillation, hemodynamic instability, and claustrophobia) had one or more coronary arterial stenoses of 50% or greater.
The prevalence of coronary arterial stenosis of 50% or greater in all the excluded patients was 1.6 (18 stenoses in 11 patients), which is similar to the prevalence of 1.7 observed in the 39 patients who completed MR coronary angiography (67 stenoses in 39 patients). Thus, this exclusion does not weaken the robustness of our results.
Currently, the two most promising noninvasive modalities for imaging coronary arteries are electron-beam computed tomography (CT) and MR imaging. The former, allowing high spatial and temporal resolution, produced good results in the first clinical applications (5,37–39). However, problems of image interpretation (5) and the high costs of the electron-beam CT units (still available in only a very few research institutions) continue to limit its use. Moreover, since the first results were published in 1995, its technology has been relatively stable (4).
After the early studies of MR imaging of coronary arteries with conventional spin-echo sequences (14) or cine MR techniques (40,41), electrocardiographically gated, fat-saturated, two-dimensional, gradient-echo, segmented–k-space, breath-hold techniques produced good results, especially for the proximal segments of the coronary arterial tree (2,3,15–25).
Manning et al (18) report 90% (44 of 49) specificity and 92% (90 of 98) sensitivity in identifying stenoses of at least 50% narrowing of the proximal segments and midportions of the major epicardial vessels. Duerinckx and Urman (2) report an overall sensitivity of 63% (17 of 27), which ranged from 73% (eight of 11) for the left anterior descending to 0% (zero of one) for the left circumflex coronary artery. Pennel et al (22) report a sensitivity of 85% (47 of 55) in 39 patients. In a similar study on 36 patients, Yoshino et al (24) obtained 88% (23 of 26) sensitivity and 98% (58 of 59) specificity. In the experience of Post et al (3), global 40% sensitivity and 97% specificity in detecting stenoses of at least 50% narrowing of the proximal segments of the coronary arteries were reported and ranged from 75% and 97% for the left main to 0% and 97% for the left circumflex coronary artery.
It should be noted that two-dimensional, gradient-echo, breath-hold MR coronary angiography has several drawbacks: (a) Relatively thick sections (4–6 mm in most studies) and gross in-plane resolution may preclude an exact grading of focal stenoses (3,5,18,22,27) and reliable identification of thin and tortuous distal vessel segments, especially of the left circumflex coronary arterial segment and collateral branches (3,5,21,22,24,25); (b) different oblique imaging planes must be selected for each artery and for each patient (21,26); (c) multiple (30 or more) 10–20-second breath holds are required, which can be a real problem in uncooperative patients (2,3,11,13,28,29); (d) inconsistent breath holding and misregistration between breath holds may lead to inadequate vessel assessment (2,11,24,26); and (e) a long examination time is required, frequently more than 60 minutes (2,19).
Experiences with electrocardiographically triggered, 3D, gradient-echo, non–breath-hold techniques with a short echo time, thin sections, and a high signal-to-noise ratio provide interesting results in identifying distal coronary vessels (42). A striking advantage of 3D coronary arterial imaging is the possibility of postprocessing and displaying coronary arteries in different planes (13). Nevertheless, an important limitation of non–breath-hold 3D acquisitions is image blurring due to respiratory motion (43).
In the past few years, the 3D, navigator-echo sequence has been proposed for coronary arterial disease assessment (12,30–33). This method, which we evaluated further in this article, is less operator dependent: No specific knowledge of coronary arterial anatomy is required before acquisition, being that a whole 3D volume is acquired during each sequence (31,44). Moreover, it does not require breath holding: Data, acquired during continuous breathing, are used for image reconstruction only if obtained in a defined range of respiratory motion (12,44); less examination time is needed in comparison with breath-hold MR angiography (31,44). Such a sequence is therefore better tolerated by the patient.
The 3D, navigator-echo sequence provides a high spatial resolution: Despite an in-plane resolution lower than that for two-dimensional breath-hold imaging, the use of 2-mm partitions allows a smaller voxel size. Higher image resolution is effective in reducing the signal intensity decrease caused by flow turbulence produced by tortuous, irregular, or stenotic arteries, with a correct assessment of the degree of stenosis. Moreover, the use of short echo times minimizes intravoxel dephasing due to blood turbulence distal to parietal abnormalities.
With the 3D, navigator-echo sequence, Li et al (12) obtained excellent results in also identifying distal branches of the left anterior descending (mean length, 11.59 cm), left circumflex (9.72 cm), and right coronary (12.59 cm) arteries. Woodard et al (29) reached a 73% sensitivity in identifying stenoses of proximal coronary arterial segments on projection images in 10 patients.
Our choice of body coil was because of the nonavailability of phased-array hardware; on the other hand, we wanted to avoid the reduction in the signal-to-noise ratio, reported by many authors using the surface coil, from the posterior vessels such as the left circumflex vessels(2,3,17,18,22). It is probable that phased-array coil receivers could enhance the signal-to-noise ratio over wider fields of view (5,12,45), even though in recent experiences with 3D navigator echo with a phased-array body coil some problems still existed in the identification of distal segments, in particular in the left circumflex coronary artery (29,34).
One of the most important advantages of 3D navigator echo, compared with two-dimensional breath-hold techniques, is the faster acquisition time (30 minutes in our protocol). On the other hand, one important drawback is a long postprocessing time, which ranges in previous reports from 30 minutes (12) to approximately 60 minutes (29). In our experience, the mean postprocessing time was about 45 minutes, with a learning curve from the first cases (more than an hour) to the last cases (about half an hour). We performed multiplanar reconstruction immediately after data acquisition: As suggested by Woodard et al (29), the possibility of evaluating multiplanar reformatted images during the same imaging session may help to reduce false-positive results, which thus increases sensitivity.
An important limitation of our study was that we did not use curved planes in data reconstruction because of the nonavailability of this software license on our equipment at the time the work was carried out. It is probable that the use of curved planes would have raised the detection rate in the distal segments, as well as the sensitivity, specificity, and predictive values for stenoses, in particular for those in the left circumflex coronary artery (4). Moreover, the use of 3D surface-rendering software may also have improved image quality and stenosis evaluation (44,46).
Technical improvements in 3D, navigator-echo MR coronary angiography are expected in the near future. As suggested by Post et al (44), the use of a too-wide acquisition window (260 msec in their experience, 178 msec in ours) is one of the reasons for the still relatively poor performances of 3D navigator echo. Woodard et al (29) demonstrated that a narrower window reduces respiratory motion artifacts, with less image blurring. k-space segmentation could improve performance, but with prolonged acquisition time; stronger and faster gradients will markedly enhance the results of 3D navigator echo (5,29).
The use of contrast agents will be an interesting field of MR coronary angiographic investigation (5,21,31,47). In a preliminary report by our group (48), we demonstrated the possibility of obtaining vascular enhancement with better visualization of distal coronary arterial segments by using 3D navigator echo with slow, intravenous, paramagnetic contrast agent administration. Better results can be achieved by using blood-pool paramagnetic contrast agents (21, 47,49).
To summarize, 3D, navigator-echo MR coronary angiography must be considered a promising sequence for the noninvasive assessment of coronary arterial stenoses. Nevertheless, if the clinical goal is to propose MR coronary angiography as a screening examination for detecting coronary arterial disease in asymptomatic patients or as a diagnostic tool in evaluating the symptoms of patients and in assessing the coronary arteries after therapeutic procedures, much work still has to be done to obtain results comparable to those provided by conventional coronary angiography.
|
Footnotes |
|---|
Abbreviation: 3D = three-dimensional
Author contributions: Guarantor of integrity of entire study, F.S.; study concepts and design, F.S., G.M.; definition of intellectual content, F.S., G.M., F.Z.; literature research, F.Z.; clinical studies, all authors; data acquisition, all authors; data analysis, F.S., F.Z.; statistical analysis, F.S., F.Z.; manuscript preparation and review, F.S., F.Z.; manuscript editing, F.Z.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
50% in diameter in native coronary arteries and in aortocoronary bypass conduits. Am J Cardiol 1997; 80:989-993.[Medline]
Related Article
This article has been cited by other articles:
|
|
 |
|
  D. A. Bluemke, S. Achenbach, M. Budoff, T. C. Gerber, B. Gersh, L. D. Hillis, W. G. Hundley, W. J. Manning, B. F. Printz, M. Stuber, et al. Noninvasive Coronary Artery Imaging: Magnetic Resonance Angiography and Multidetector Computed Tomography Angiography: A Scientific Statement From the American Heart Association Committee on Cardiovascular Imaging and Intervention of the Council on Cardiovascular Radiology and Intervention, and the Councils on Clinical Cardiology and Cardiovascular Disease in the Young Circulation, July 29, 2008; 118(5): 586 - 606. [Full Text] [PDF]
|
|
|
|
 |
|
 T. H. Hauser and W. J. Manning Coronary MRI: More Pretty Pictures or Present-Day Value? J. Am. Coll. Cardiol., November 21, 2006; 48(10): 1951 - 1952. [Full Text] [PDF]
|
|
|
|
|
|
P. G. Danias, A. Roussakis, and J. P.A. Ioannidis Diagnostic performance of coronary magnetic resonance angiography as compared against conventional x-ray angiography: A meta-analysis J. Am. Coll. Cardiol., November 2, 2004; 44(9): 1867 - 1876. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Dhawan, K. C. Dharmashankar, and T. Tak Role of Magnetic Resonance Imaging in Visualizing Coronary Arteries Clin. Med. Res., August 1, 2004; 2(3): 173 - 179. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. S. Dirksen, H. J. Lamb, P. Kunz, P. Robert, C. Corot, and A. de Roos Improved MR Coronary Angiography with Use of a New Rapid Clearance Blood Pool Contrast Agent in Pigs Radiology, June 1, 2003; 227(3): 802 - 808. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Bogaert, R. Kuzo, S. Dymarkowski, R. Beckers, J. Piessens, and F. E. Rademakers Coronary Artery Imaging with Real-time Navigator Three-dimensional Turbo-Field-Echo MR Coronary Angiography: Initial Experience Radiology, March 1, 2003; 226(3): 707 - 716. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. A. Fayad, V. Fuster, K. Nikolaou, and C. Becker Computed Tomography and Magnetic Resonance Imaging for Noninvasive Coronary Angiography and Plaque Imaging: Current and Potential Future Concepts Circulation, October 8, 2002; 106(15): 2026 - 2034. [Full Text] [PDF]
|
|
|
|
|
|
J. F. Polak MR Coronary Angiography: Are We There Yet? Radiology, March 1, 2000; 214(3): 649 - 650. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
