| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Vascular and Interventional Radiology |
1 From the Departments of Radiology (F.S.P., J.D.C., J.C.C., C.F., R.M.M., G.M.B.) and Vascular Surgery (M.D.M.), Feinberg School of Medicine, Northwestern University, Chicago, Ill; Department of Radiology, Arizona Health Sciences Center, Tucson, Ariz (E.A.K.); and Department of Radiology, Geffen School of Medicine, UCLA, Los Angeles, Calif (J.P.F.). Received July 19, 2004; revision requested September 2; revision received January 12, 2005; final version accepted December 8. Address correspondence to F.S.P., Salinas Valley Radiologists, 559 Abbott St, Salinas, CA 93901 (e-mail: s-pereles{at}northwestern.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Materials and Methods: This retrospective study protocol received approval from the Office of Sponsored Research at Northwestern University, which included review by the Office for the Protection of Research Subjects. Informed consent was waived for this HIPAA-compliant study. One hundred twenty-one consecutive patients (67 men: mean age, 66 years ± 12 [standard deviation]; 54 women: mean age, 69 years ± 14), who were referred for evaluation of peripheral vascular disease, underwent peripheral contrast material–enhanced MR angiography. By using a hybrid technique, two independent timing measurements were performed in the pelvis and calves followed by MR angiography of the calves and, subsequently, a pelvis-thigh stepping-table acquisition. Images were evaluated for extent of disease, on the basis of degree of stenosis; for venous contamination, on the basis of venous signal intensity; and for diagnostic quality, on the basis of diagnostic confidence of the observer. DSA correlation of the extent of vascular disease was available in 45 of these patients, which was used to evaluate the diagnostic power of the hybrid technique.
Results: For detection of stenosis greater than 50%, the hybrid technique had 95% sensitivity (P < .05), 95% specificity (P < .05), and 95% accuracy (P < .05). There was no significant venous contamination in any of the examinations performed with this technique.
Conclusion: The hybrid peripheral MR angiography technique provides diagnostic-quality examinations and virtually eliminates venous contamination.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Several approaches have been taken to improve quality and reproducibility of lower extremity contrast-enhanced MR angiography, hereafter referred to as peripheral MR angiography, while decreasing examination times. Early studies utilized the main body coil for signal reception, but more recent studies have shown that the use of surface coils is preferable to signal reception from the main body coil alone (4,5). Multistation studies to cover the peripheral vascular runoff territories from the renal arteries to the calves were originally performed by moving a single phased-array coil from station to station for separate acquisitions of the calves, thighs, and pelvis. Separate contrast medium injections were performed for each imaging station (6). Timing of image acquisitions was either a best-guess technique or based on a single contrast medium timing run from the antecubital venous injection site to the lower abdominal or pelvic arteries. Contrast medium injections and timing for thigh and calf imaging stations were prescribed by estimating the additional transit times from the pelvis to these stations. Subsequent studies have demonstrated that transit times from antecubital venous injection to the feet are highly variable (7,8). This variability in timing can adversely affect image quality with previous peripheral MR angiographic acquisition techniques.
The advent of manual movement and automated stepping-table techniques, coupled with movable array coils or dedicated peripheral array coils, respectively, further improved the speed of peripheral MR angiography by capitalizing on bolus-chase methods of contrast material infusion (5,9,10). However, imaging the pelvis, thighs, and calves by chasing the arterial bolus to the feet before venous contamination occurs can be a formidable challenge. While parallel imaging methods such as sensitivity encoding, or SENSE, have increased the speed of MR angiography and innovative k-space sampling techniques such as elliptical centric k-space acquisition and projection reconstruction undersampling like time-resolved imaging of contrast kinetics, or PR TRICKS, have improved peripheral MR angiography, no one technique is the universal standard (3,11–14).
We hypothesized that a more universally applicable contrast material injection and acquisition scheme is required to satisfy the rigorous demands of vascular surgeons in employing peripheral MR angiography as a substitute for catheter-based DSA. In performing peripheral MR angiography, the following traits would be most desirable for optimal examinations: First, the technique would have to achieve near-isotropic spatial resolution, with maximal voxel dimensions approaching 1 mm in the small vessels of the calves and proximal foot, where evaluation of graft touchdown sites is critical. Second, the technique would have to preserve both in-plane and through-plane spatial resolution in the pelvis and thigh stations for adequate three-dimensional (3D) reconstruction. This means that in an effort to avoid venous contamination, employing low through-plane resolution, projectional techniques to cut imaging times of the pelvis and thighs is not an acceptable solution. Projectional techniques do not preserve adequate spatial resolution to generate diagnostic 3D reconstructions. Third, the examination coverage should be from the renal arteries to the pedal arch at a minimum. This coverage thereby includes important additional information about coincident renal disease in patients with peripheral arterial occlusive disease. Also demonstrated is the most distal pedal disease, often missed by means of previous techniques but of clinical interest to vascular surgeons and radiologists when planning therapeutic interventions. Finally, the technique must be free of venous contamination to allow accurate interpretation and avoid the possibility of calling patients back for additional MR imaging at a second appointment.
We developed a technique of peripheral MR angiography that employs dual timing acquisitions in both the pelvis and calves to avoid timing errors in the most distal station (7). This "hybrid" approach, as we call it, synthesizes multistation multi-injection techniques with bolus-chase methods. In addition, acquisitions of the feet are obtained before imaging the pelvis and thighs to minimize venous contamination and background signal. Repeated acquisitions are obtained in the distal station to ensure optimal imaging of each leg in cases where transit times to one foot are markedly different than those to the other, such as in patients with asymmetric disease or where bypass grafts are in place. This information is known as a result of the added calf timing run prior to 3D acquisition. The pelvis and thigh stepping-table images are obtained after the calf images and therefore acquisition can be slightly longer than with pure bolus-chase methods because the feet and calves have already been imaged. This extra time permits acquisition of increased spatial resolution at each station without risk of venous contamination. The purpose of our study, therefore, was to retrospectively evaluate the accuracy of hybrid peripheral MR angiography by using DSA as the reference standard.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The study protocol received approval from the Office of Sponsored Research (at Northwestern University), which included review by the Office for the Protection of Research Subjects. A waiver of informed consent was obtained for our retrospective Health Insurance Portability and Accountability Act–compliant study. Patients provided informed consent before being evaluated with the hybrid MR angiography technique used in our study.
Hybrid MR Angiography Technique
Patients were imaged with the hybrid technique by means of a 3D spoiled gradient echo pulse sequence with a 1.5-T MR imager (Magnetom Quantum; Siemens Medical Solutions, Malvern, Pa) with a gradient of 30 mT/m and a slew rate of 100 (mT · m–1)/msec. A dedicated peripheral vascular coil (Siemens Medical Solutions) was also used. In-plane resolution was on the order of 0.7 x 0.9 mm. Through-plane resolution was between 0.9 and 1.2 mm depending on body habitus, with an 80-mm slab covering from the posterior arteries of the foot to the distal metatarsals and plantar arch. Imaging parameters were as described in Table 1.
|
Two milliliters of gadopentetate dimeglumine (Magnevist; Berlex Pharmaceutical, Wayne, NJ) was injected for the timing run at 2 mL/sec followed by a 20-mL saline bolus at the same rate by using a power injector (Medrad, Indianola, Pa). Next the pelvis timing run was performed by using a 2-mL injection of gadolinium-based contrast material followed by a 20-mL saline flush; both were administered with a power injector at 2 mL/sec. The pelvis timing run was performed before 3D MR angiography of the calves to avoid having the pelvis timing run obscured by the 20-mL bolus used to image the calves. Additionally, in the event that vascular disease in the calves was so severe that the vessels were occluded at the level of the calf timing run, the pelvis timing could serve as a reference point to help estimate the appropriate acquisition start time for the calves.
A precontrast mask acquisition was obtained in the calves and feet. Onset of calf station contrast-enhanced MR angiographic acquisition was based on the first arrival of contrast material demonstrated at the calf timing run. A 20-mL contrast material bolus at 2 mL/sec was infused with two consecutive acquisitions obtained with no delay between acquisitions. Hybrid calf acquisitions averaged 22 seconds ± 3. If contrast material arrival in each calf was demonstrated to be discrepant by more than 10 seconds on the calf timing run, then a third consecutive acquisition was obtained at this station.
Following imaging of the calves and feet station, mask acquisitions for the pelvis and thigh stations were obtained. A second contrast material infusion was performed with automated stepping-table method contrast-enhanced MR angiography of the pelvis and thighs. The start of the pelvic acquisition was derived from the pelvic timing run. The thigh acquisition immediately followed the pelvis acquisition by using the stepping-table method in a manner similar to bolus-chase techniques. The contrast material injection was performed with a 36-mL bolus injected at a divided rate. The first 20 mL of contrast material was injected at 2 mL/sec, followed by the remaining 16 mL at 0.8 mL/sec (total pelvis and thigh contrast material infusion time, 30 seconds), and then a 20-mL saline flush at 2 mL/sec. The parameters (eg, field of view, matrix size, number of partitions) at each station were optimized to patient body habitus to permit the most rapid imaging possible without clinically compromising through-plane resolution (Table 1).
Average imaging times for each station were 15 seconds ± 3 for the pelvis and 14 seconds ± 3 for the thighs. Table movement time between pelvis and thigh stations was 7 seconds, which makes the average total pelvis-thigh acquisition time 36 seconds. For any patients in whom 60 mL of contrast medium exceeded a dose of 0.3 mmol per kilogram of body weight, the dose was reduced appropriately so that the calf station was still performed with 20 mL and the pelvis-thigh contrast medium infusion was reduced to a remaining cumulative dose of 0.3 mmol/kg. Infusion remained at divided rates of 20 mL at 2 mL/sec followed by the remaining contrast material at 0.8 mL/sec. In the most extreme three cases, where, for example, one patient's weight was only 50 kg, infusion was adjusted accordingly to a 14-mL calf infusion with both timing run and actual acquisition infusions at 1.5 mL/sec. In this same patient, pelvis and calf timing runs were reduced to 1-mL volume infused at 1.5 mL/sec. Also, a 15-mL pelvis-thigh infusion was infused as 10 mL at 1.5 mL/sec followed by 5 mL at 0.5 mL/sec.
MR Angiogram Analysis
One board-certified radiologist specializing in cardiovascular imaging (F.S.P.) and one vascular surgeon (M.D.M.), with 6 and 10 years of cardiovascular imaging experience, respectively, performed the image evaluations. Blinded to patient identity and clinical history, they independently evaluated two-dimensional source partitions, technologist-generated maximum intensity projection, and volume-rendered 3D MR images for presence of disease at each of 29 anatomic segments in 121 cases by using a method previously described in the vascular surgery literature (15). All images were evaluated as soft-copy images on picture archiving and communication system workstation monitors (Centricity; GE Medical Systems, Milwaukee, Wis), and 3D volume renderings were freely manipulable by the evaluators on adjacent 3D workstations (Vitrea; Vital Images, Plymouth, Minn). Consensus was obtained after image analysis to arrive at final values.
Disease was graded into five separate categories as occluded (100% stenosis), severe (76%–99% stenosis), moderate (51%–75% stenosis), mild (26%–50% stenosis), and not significant (0%–25% stenosis). The most severe focus of disease was graded in each segment.
Venous contamination was graded at each of the three MR imaging stations for all 121 subjects on a four-point scale: 1, not visible; 2, barely visible but does not affect diagnostic interpretation; 3, visible and may affect diagnostic interpretation; and 4, present and definitely compromises diagnostic interpretation.
Diagnostic quality of images was also graded for all 121 subjects at each of the three imaging stations on a scale of 1–5: 1, nondiagnostic; 2, poor quality and observer not confident; 3, fair quality and observer marginally confident; 4, good quality and observer confident; and 5, excellent quality and observer highly confident. P values of less than .05 were considered to indicate a significant difference.
DSA Technique and Image Analysis
Forty-five patients underwent DSA for correlation in 763 segments. The total does not equal 1305 segments because some patients had undergone amputation and some angiograms were obtained with selective unilateral injections. Vessels imaged at DSA were those deemed necessary during a therapeutic interventional procedure or when MR angiograms revealed pathologic conditions that warranted further investigation before these procedures. All DSA examinations were performed within 1 month of the corresponding MR angiography examination. The DSA technique utilized in this study is the same as that described by Swan et al (3). In brief, a Seldinger approach was used to introduce a 5-F angiographic catheter into the common femoral artery. Injections were performed by using iohexol (Omnipaque; Amersham Health, Princeton, NJ), a nonionic contrast agent, or iodixanol (Visipaque; Amersham Health) in cases of creatinine levels greater than 2.0. DSA images were acquired with the Integris angiographic unit (Philips, Best, the Netherlands) by using a 17-inch image intensifier for a 38-cm field of view and a 1024 x 1024 matrix. Contrast agent dose per injection varied from 12 to 20 mL injected at a rate of 6–10 mL depending on the vessel being injected and imaged. The DSA images were read by a board-certified vascular surgeon (M.D.M.) with 10 years of experience in vascular imaging.
For statistical analysis, DSA images were graded for disease on the basis of the same five-category scale used for peripheral MR angiography. DSA images were analyzed 6 months after MR angiogram analysis to avoid recall bias.
Statistical Analysis
Statistical analysis was performed to assess the grading of disease in the 763 segments for which both MR angiography and DSA had been performed. By using corresponding DSA segments as a reference standard, the hybrid technique was evaluated for its sensitivity, specificity, and accuracy in depicting vascular disease. For both MR angiography and DSA, segments showing moderate or severe stenosis were considered diseased (>50% stenosis). Two-tailed Student t tests for data sets with unequal variance were used to evaluate these data, and P values of less than .01 were considered indicative of statistically significant differences. All calculations were performed with commercially available software (Excel 2002; Microsoft, Redmond, Wash).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
In order to give an overall picture of the technique, we calculated these values across all three stations (pelvis, thigh, and calf) by using the patient as the unit of analysis. Under this assumption, if at least one of the 29 anatomic segments evaluated was found to be positive for stenosis, the patient was considered to have a positive case for disease. This binary classification, although it results in less statistical power, avoids a data clustering problem that arises in calculating the sensitivity and specificity by using individual segments or regions, which are not truly independent variables—segments or regions within a single patient may be more alike than segments or regions in other patients.
The average number of segments per patient was 6 (standard deviation, 4.14; range, 1–16). Of the 45 patients with matching segments, 44 had at least one positive segment that was read as positive on both types of images, resulting in a per-patient sensitivity of 98%. The one patient with false-negative findings had only one segment that was called positive at DSA and that was not diagnosed at MR angiography. For the 45 patients with disease, there were 16 patients who had at least one segment called false-positive at MR angiography compared with DSA, so the specificity on a per-patient basis was 64%. An important caveat is that after blinded reviews and statistical analysis, an informal comparison of these cases showed that findings at MR angiography were correct in nearly every case. Despite this fact, DSA is the reference standard of this study, and so these cases were still categorized as "false-positive."
Sensitivity, specificity, and accuracy were also calculated for each of the 29 individual segments (Table 2). This analysis assumes the segments are independent variables and was performed to get a broader picture of the degree to which each stenosis detected on the DSA images was also detected on the MR angiograms and exactly how many individual false-positive findings occurred on the MR angiograms. Notably, the pelvis, thigh, and calf stations all had similarly high values for sensitivity, specificity, and accuracy. Accuracy was high at all three stations but most notable was the high level of accuracy at the calf station, where vessels are the smallest and typically the most difficult to image.
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The hybrid technique was born out of 3 years of attempts using our equipment to replicate and improve upon others' techniques and results obtained on a variety of vendors' equipment. Technologic advancements in both hardware and software over the past 10 years have improved speed of acquisition, increased spatial resolution per equivalent time duration, and increased signal-to-noise ratio. At the time of this writing, despite the development of improved hardware with "ultrafast" gradients (repetition times less than 2 msec), signal-to-noise ratio improvements from peripheral phased-array surface coils, and employment of acquisition acceleration factors such as parallel imaging techniques and k-space undersampling, a relatively immutable fact has emerged: Venous overlay in the calves and feet has remained a problem in at least a small percentage (5%–10%) of cases (15). To that end, multiple groups have tried techniques that increase the speed of acquisition and table movement so that imaging of the feet can occur as early as possible after initiation of contrast material infusion. Work by independent groups has demonstrated that transit times for contrast material to travel from antecubital venous infusion to the calves and feet can occur in adults in less than 15 seconds and that venous return in the calves can occur in as quickly as 18 seconds (7,8).
To put this in perspective, let us assume that as of the year 2005, the fastest commercial 1.5-T MR imagers are capable of acquiring adequate 3D MR angiographic volumes of the pelvis and thighs in approximately 10 seconds, each with a matrix of 512 (or better), with a through-plane resolution of no more than 2 mm, and employing a parallel imaging acceleration factor of two. Automated table movement remains, at best, 5 seconds per interstation travel (pelvis to thigh and thigh to calf) for a total of 10 seconds of table movement. Therefore the fastest bolus-chase peripheral MR angiography techniques would require a minimum of 30 seconds before the initiation of calf imaging can occur. In our experience this still results in an approximately 5%–10% venous contamination rate (15). Furthermore, machines without the latest hardware and software could not achieve these parameters and would likely have higher venous contamination rates.
Recently, application of tourniquets or blood pressure cuffs in a suprapopliteal location has been employed by some groups in efforts to retard venous return and decrease venous overlay (16). This practice, although relatively effective, is not well tolerated by some patients. In people with the most severe peripheral vascular disease, such as patients with rest pain and/or arterial insufficiency ulcers, just placing a peripheral vascular coil and lying on the table for imaging purposes is often the maximum that they can bear. Justified or not, many technologists dislike the thought of having to be responsible for checking a patient's blood pressure prior to an MR examination and further dislike having to accurately place and inflate bilateral blood pressure cuffs to venous occlusive pressures of 40–60 mm Hg.
A common sense approach to this problem is to use a hybrid approach and acquire images at the most distal (calf and foot) imaging station first to avoid venous contamination. This method was first facilitated with software algorithms that permit in-line subtracted timing runs of the small vessels in the calves (7). Such software is now available in some form from all of the major MR imager vendors. With a priori knowledge of the transit time of contrast material to the calves, the calf-foot acquisition station can be optimized for coverage, spatial resolution, and timing. Even the challenge of large discrepancies in flow to different legs in the same patient is overcome. With the hybrid technique, one is able to accurately acquire arterial phase images of the arrival of contrast material in the leg with faster-flowing contrast material first and then obtain the immediate serial acquisition of the calf-foot station that depicts arterial anatomy of the limb with slower flow.
Since the publication of our preliminary results in the Journal of Vascular Surgery in 2003, von Kalle et al (17) have published a cohort study comparing the hybrid technique with standard bolus-chase peripheral MR angiography of the lower extremities. Their study of 80 individuals came to the conclusion that "moving table hybrid CEMRA [contrast-enhanced MR angiography] is superior to conventional technique in craniocaudal direction by producing less venous overlap of arteries and is especially more suitable for the diagnostic evaluation of the cruropedal region." The results of this study are consistent with our data that the hybrid technique is an improvement upon bolus-chase technique for the evaluation of the lower extremity vessels.
Other groups have adopted the general premise of the hybrid technique but have employed dynamic or time-resolved projectional imaging of the calves and feet (18,19). This technique provides excellent temporal flow information with good in-plane visualization of the distal vessels. We had employed this approach in the past, however, our vascular surgeons demanded higher through-plane resolution (near-isotropic, millimeter voxel size) so that they could accurately evaluate for viable graft touchdown sites. For this reason, we employ the highest spatial resolution 3D MR acquisition when imaging the feet and calf station. It produces diagnostic morphologic depiction of the distal vasculature not present with projectional techniques, while also providing temporal information from the second serial calf acquisition that would be missed by using the bolus-chase method.
A temporally resolved technique such as "TRICKS" would incorporate well with the hybrid technique at the calf station. By performing this type of high spatial and temporal isolation imaging of the calves and feet and then following with pelvis and thigh imaging, the advantages of both strategies could be obtained.
With the calf-foot station images acquired first with a separate injection of contrast material, the subsequent pelvis and thigh stepping-table runoff acquisition can be obtained with a second injection of contrast material. This technique affords greater acquisition times in both the pelvis and thigh stations without having to race to a distal calf station before there is risk of venous contamination. The result is higher spatial resolution at the proximal two stations, as well as higher spatial resolution at the distal calf station, and the risk of significant venous contamination is virtually eliminated.
Although repeated examinations in individuals by using both the standard bolus-chase technique and the hybrid technique would have been a more scientifically rigorous comparison, ethical considerations and prohibitive costs made such a study unfeasible. A further limitation of the hybrid technique is that it uses more contrast material than does the standard bolus-chase technique, increasing the cost of each lower extremity MR angiography study. However, by eliminating the repeat studies sometimes necessitated by venous contamination, the hybrid technique may reduce overall imaging costs.
The reported accuracy in the literature at the pelvis and thigh stations is high with both bolus-chase techniques but is slightly greater with the hybrid technique, which is likely owing to the higher spatial resolution obtainable (3,9,10,14).
Reduction of venous overlay and improved spatial resolution also led to overall increased diagnostic confidence with the hybrid technique. This was again most important for the calves and feet where venous contamination is most frequent.
The hybrid peripheral MR angiography procedure, based on the principle of obtaining MR angiograms of the most distal station first, followed by stepping-table MR angiograms of proximal stations, has proved to be a robust technique. The hybrid peripheral MR angiography technique provides diagnostic-quality images and virtually eliminates venous contamination. At our institution, the radiologists and vascular surgeons find the hybrid MR angiography technique of sufficient quality to replace diagnostic peripheral DSA. The hybrid technique, because it does not require specialized equipment or software, should have universal applicability regardless of one's MR vendor. Until MR machines are capable of performing bolus-chase methods with adequate spatial resolution for pelvis and thigh imaging while still being able to reach the feet in less than 20 seconds after contrast material infusion, hybrid methods should be considered for dependable performance of peripheral MR angiography. This paradigm could shift again with advances in parallel imaging capabilities on future MR machines.
|
FOOTNOTES |
|---|
Abbreviations: DSA = digital subtraction angiography • 3D = three-dimensional
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, F.S.P., M.D.M.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, F.S.P., J.D.C., J.C.C., C.F., R.M.M., G.M.B.; clinical studies, F.S.P., J.D.C., J.C.C., C.F., M.D.M., R.M.M., J.P.F.; experimental studies, F.S.P., J.D.C., J.C.C., E.A.K., J.P.F.; statistical analysis, F.S.P., E.A.K., G.M.B.; and manuscript editing, F.S.P., M.D.M., R.M.M., E.A.K., G.M.B.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Habibi, M. S. Krishnam, D. G. Lohan, F. Barkhordarian, M. Jalili, R. S. Saleh, S. G. Ruehm, and J. P. Finn High-Spatial-Resolution Lower Extremity MR Angiography at 3.0 T: Contrast Agent Dose Comparison Study Radiology, August 1, 2008; 248(2): 680 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Ersoy and F. J. Rybicki MR Angiography of the Lower Extremities Am. J. Roentgenol., June 1, 2008; 190(6): 1675 - 1684. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. G. Lohan, A. Tomasian, M. Krishnam, P. Jonnala, K. E. Blackwell, and J. P. Finn MR Angiography of Lower Extremities at 3 T: Presurgical Planning of Fibular Free Flap Transfer for Facial Reconstruction Am. J. Roentgenol., March 1, 2008; 190(3): 770 - 776. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Sabharwal, N. Fotiadis, and A. Adam Modern trends in interventional radiology Br. Med. Bull., April 30, 2007; (2007) ldm006v1. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
