| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vascular and Interventional Radiology |
1 From the Departments of Academic Radiology (D.G.W.F., A.R.M., P.S.M.), Radiology (I.R.D.), and Medical Physics (A.L.M.), Queen’s Medical Centre, Nottingham, England. Received January 28, 2002; revision requested March 13; revision received June 12; accepted July 25. Supported by grant RB2305 from the British Heart Foundation. Address correspondence to A.R.M., Department of Medical Imaging, Sunnybrook and Women’s Health Sciences Centre, 2075 Bayview Ave, Toronto M4N 3M5, Ontario, Canada (e-mail: alan.moody@swchsc.on.ca).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: A single anteroposterior maximum intensity projection (MIP) venogram of the femoral and iliac veins was constructed by using VESPA MR venography in 55 symptomatic patients suspected of having lower limb DVT. All patients also underwent conventional venography, results of which were used as the standard of reference. VESPA MR venograms were interpreted by two independent reviewers (reviewers A and B) who were unaware of other results. Sensitivity and specificity of VESPA MR venography for the diagnosis of thrombus in the femoral and iliac veins were calculated. Interobserver variability was calculated for these observations by using weighted 
with equally spaced weights for positive, nondiagnostic, and negative studies. Nondiagnostic studies were reinterpreted separately by reviewer A on the basis of source data.
RESULTS: Sensitivity of VESPA MR venography for the femoral veins (20 of 20) and iliac veins (seven of seven) was 100% for both reviewers. Specificity was 100% (39 of 39 for reviewer A, 40 of 40 for reviewer B) for the iliac veins and 97% (31 of 32) for the femoral veins for both reviewers. Segments in which the VESPA MR venograms were nondiagnostic were excluded from this analysis. Interobserver variability as calculated by using weighted for positive, negative, and nondiagnostic studies was 0.85 for femoral veins and 0.97 for iliac veins. Interpretation of the source data led to correct diagnosis in six of six cases in which the VESPA MR venograms were nondiagnostic.
CONCLUSION: VESPA MR venography yielded MIP venograms that were highly accurate for the diagnosis of DVT in femoral and iliac veins. Interpretation of the studies was also highly reproducible.
© RSNA, 2003
Index terms: Magnetic resonance (MR), image processing • Veins, extremities, 938.751, 988.751 • Veins, MR, 938.12942, 988.12942 • Veins, thrombosis, 938.751, 988.751
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Contrast material–enhanced MR venography involves the use of the same rapid, three-dimensional (3D) sequences that have been developed for arteriography. The problems of time-of-flight techniques are overcome with contrast-enhanced MR venography, but images obtained at contrast-enhanced MR venography require postprocessing so that arterial signal is removed. With this technique, contrast material is injected into a peripheral vein and serial image volume measurements are acquired. Subtraction of an early arterial phase measurement from a late arterial-venous equilibrium phase measurement yields a selective venous angiogram (8).
In a previous study, we demonstrated that a double-subtraction algorithm, in which two early and two late measurements are used, yields images of higher venous signal intensity than does a single-subtraction algorithm and enables MIP venograms of high diagnostic quality to be obtained (9). We have described this technique as venous enhanced subtracted peak arterial (VESPA) MR venography. Thus, the purpose of this study was to assess diagnostic accuracy and interobserver variability at VESPA MR venography compared with those at conventional venography in the diagnosis of deep venous thrombosis (DVT) in the femoral and iliac veins.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
We did not recruit consecutive patients owing to difficulties involving access to MR imaging sessions. However, this study was prospective and blinded, and patients were not selected with regard to the extent of thrombosis. During the study period (July 1998 to September 1999), 247 conventional venographic studies (of which 46 were interpreted as positive for proximal thrombosis) were performed in a pool of patients from which the study patients were selected. Patient exclusion criteria were the following: contraindications to MR imaging, claustrophobia, inability to lie flat, and inconclusive venographic results.
MR Imaging
MR images were acquired with a 1.5-T MR imaging unit (Magnetom Vision; Siemens, Erlangen, Germany) and a body coil by using a 3D gradient-echo sequence (repetition time msec/echo time msec, 5/2; flip angle, 35°; 256 x 200 matrix with frequency encoding perpendicular to the transverse plane; 30 partitions of 3.5 mm). A single field of view (500 mm) included the lower 5–10 cm of the IVC and the iliac and femoral veins. Eight successive image acquisitions, which lasted 30 seconds each, were performed. Twenty milligrams of contrast material (gadopentetate dimeglumine, Magnevist; Schering, Berlin, Germany) was injected at approximately 1 mL per second into the antecubital fossa. This injection was performed after 15 seconds of the first acquisition had elapsed and was followed by a 20-mL saline flush. MR imaging was well tolerated in all cases.
A vacuum fixation bean bag (Qados, Camberley, England) was used to minimize leg movement. This 60 x 90-cm bag was placed distally under each patient’s knees, calves, and feet, and, following suction, it became rigidly molded to the patient’s legs, fixing them in position.
Postprocessing of MR Images
A double-subtraction algorithm was used (Fig 1). In most cases this involved subtraction of the sum of the first two measurements (1 and 2) from the sum of measurements 7 and 8 (Fig 1). We have shown previously that this process achieves subtraction of both arterial and background signal while retaining the summed venous signal from measurements 7 and 8 (8). For two patients, however, this subtraction did not remove the arterial signal because prolonged circulation time in these patients resulted in a delay in the arterial phase. When subtraction of measurements 1 and 3 from measurements 7 and 8 was performed for the data of these two patients, arterial signal was successfully removed. In this way, accountability for the effect of different circulation times was achieved without the need for timing studies. A single anteroposterior MIP venogram was produced from the processed data set for each patient. Imaging time was 4 minutes in all cases, and combined image acquisition and postprocessing time was less than 10 minutes in all cases.
|
Interpretation of Studies
Findings at VESPA MR venography in each patient were interpreted without reference to source data by two radiologists (A.R.M. [reviewer A] and I.R.D. [reviewer B]) on the basis of a single anteroposterior MIP of the processed data set. Radiologist A was an MR imaging and general radiologist with 12 years of reading experience, and reviewer B was a vascular and general radiologist with 8 years of reading experience. The radiologists were unaware of each other’s results and the results of conventional venography. Although both legs were simultaneously imaged in all patients, only the veins on the symptomatic side were evaluated. The femoral and iliac venous segments were evaluated separately. A study was considered positive for DVT if intraluminal filling defects were seen or if nonfilling of veins with a sharp cutoff was detected (Fig 2). A study was considered nondiagnostic if the venous filling was indistinct without a clear cutoff or a clear filling defect. Reviewer A also recorded the proximal extent of thrombosis.
|
The proximal extent of thrombosis as recorded by reviewer A on the basis of VESPA MR venographic findings was compared with the proximal extent as recorded by the duty radiologist on the basis of conventional venographic findings. This evaluation was performed by comparing the most proximal venous segment that had been considered to contain thrombus at each modality. Venous segments were defined as femoral (ie, the common, deep, and superficial femoral veins), iliac (ie, the common and external iliac veins), and the IVC. Comparison of the extent of thrombosis within these segments was not performed. Comparison of the results from conventional and VESPA MR venography was then performed by one author (D.G.W.F.) not involved in the interpretation of either set of images.
At the end of this investigation, reviewer A, who was not aware of the results of conventional venography or patient details, reinterpreted the nondiagnostic VESPA MR studies with reference to source data but not the MIP venograms. After this, reviewer A performed unblinded analysis of both the nondiagnostic and the false-positive MIP VESPA MR venograms in conjunction with analysis of source data and the duty radiologists’ initial interpretations of conventional venograms to determine the reasons for the diagnostic failure.
Statistical Analysis
Sensitivity and specificity of VESPA MR venography for the diagnosis of thrombus in the femoral and iliac veins were calculated. Nondiagnostic cases were not included in this analysis. Interobserver error was calculated for these observations by using the weighted statistic with equally spaced weights for positive, nondiagnostic, and negative studies. Calculations were performed by using the Statistical Package for Social Sciences software (SPSS, Chicago, Ill).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
for positive, nondiagnostic, and negative segments, was 0.85 for the femoral veins and 0.97 for the iliac veins.
|
|
|
Proximal Extent of Thrombosis
Results of conventional venography were nondiagnostic because of poor filling in the iliac veins and the IVC in eight of 13 patients with femoral DVT. In addition, results of conventional venography were nondiagnostic for the IVC in six of seven patients with iliofemoral DVT. Conventional venography therefore failed to depict the proximal extent of thrombosis in 14 of 20 patients with femoral or iliofemoral thrombosis.
VESPA MR venograms clearly depicted the proximal extent of thrombosis in all cases. Findings at VESPA MR venography concurred with those at conventional venography in the six patients in whom conventional venography had depicted the proximal extent of femoral or iliofemoral thrombosis. VESPA MR venography enabled visualization of thrombus in six of the 14 segments in which results of conventional venography were nondiagnostic. In four of these six cases, VESPA MR venography enabled visualization of unsuspected IVC thrombus (Table 3, Fig 5).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Study Strengths and Weaknesses
The VESPA technique incorporates the use of a standard angiography sequence with standard postprocessing software, and reviewer B was a vascular and general radiologist who was not experienced in MR imaging. Our results should therefore be readily reproducible in other clinical units. The ability to display the MR images as MIP venograms with suppressed arterial and background signal enabled the studies to be interpreted rapidly and easily. We did not provide source data to the reviewers so that we could test the accuracy of and potential problems associated with interpretation of a single MIP venogram. However, the use of a single MIP venogram could theoretically result in the failure to visualize nonocclusive thrombus. Nonocclusive thrombi were not encountered in this study and are uncommon in symptomatic patients, but they are more frequently found in screening of high-risk asymptomatic patients (10).
Although the study patients were not consecutive, there were representative proportions of patients with differing extents of thrombosis and an even split between inpatients and outpatients. Exclusion criteria were limited to nondiagnostic studies and contraindications to MR imaging. Patient age range and duration of symptoms were widely spread, and the study included six patients with metallic hip replacements. Reviewer A interpreted the MIP VESPA MR venograms in all six of these patients correctly; reviewer B interpreted five of six of the venograms correctly and considered one to be nondiagnostic.
Each reviewer interpreted the MIP VESPA MR venograms of three (5%) of the 55 femoral segments as nondiagnostic. One MR venogram that was considered nondiagnostic for the femoral veins was interpreted by Reviewer A as also being nondiagnostic for the iliac veins. Reinterpretation of these nondiagnostic studies based on source data led to a diagnosis that concurred with that rendered at conventional venography in all cases. Further analysis showed that irregularities of the venous outline had been either introduced or accentuated by the subtraction and MIP processes. Subtraction errors are a weakness of any technique that combines multiple acquisitions and were reduced but not eliminated by fixation of the patients’ legs with the bean bag. Postprocessing techniques that involve the use of filtering and motion detection algorithms can be used to limit the effect of such movement (11).
VESPA MR venography depicted a substantially greater portion of the proximal extent of thrombus than did conventional venography in six of the 20 patients with femoral or iliofemoral thrombosis. Thrombus within the IVC that would have been associated with a high risk of pulmonary embolism was diagnosed at VESPA MR venography and not visualized at conventional venography in four of these cases. The additional regions of thrombosis were not confirmed with further testing because it was decided that this was not clinically justified. However, these findings are in keeping with those of previous studies (as described in the next section of this article).
The single field of view (500 mm) from the lower IVC to the femoral veins used at VESPA MR venography did not cover the popliteal or calf veins, so the accuracy of the technique for the diagnosis of thrombosis in these regions was not assessed. In our experience, the popliteal veins can be imaged very successfully at MR by performing an additional distal imaging block from the calf to the lower part of the SFV. We have found that the calf veins are more difficult to image because some of the veins are in a semicollapsed state when the patient is in the supine position. Additionally, imaging of these veins is more prone to subtraction errors because of the small size of the veins. With further refinements of this technique, however, it is hoped that satisfactory imaging of the calf veins will be possible.
Comparison of the Results of This Study with Those of Other Studies
The diagnostic accuracy of time-of-flight MR venography has been assessed in several studies. To our knowledge, this is the first study to assess the accuracy of contrast-enhanced MR venography for the diagnosis of DVT of the lower limb. The VESPA technique involves a double-subtraction algorithm in which two early and two delayed measurements are used. Both single-subtraction and factor analysis algorithms have been used to remove arterial and background signal, but the diagnostic accuracy of these techniques has not been assessed. Use of the VESPA double-subtraction algorithm increases venous signal compared with use of a single subtraction algorithm and does not require the identification of arterial and venous segments and the additional software that is required for factor analysis (9,12).
Although the use of contrast material adds to the cost of the procedure, this is offset by a more rapid imaging time than is achievable with time-of-flight MR imaging. Contrast-enhanced MR venography benefits from the use of high-spatial-resolution 3D sequences that enable complete visualization of the venous anatomy and allow reconstruction techniques to be used. Because venous signal is not derived from the motion of blood, the technique is not impaired by motion artifacts. Although subtraction and other artifacts did occur, we found that these artifacts usually led to nondiagnostic rather than to false-positive studies, and analysis of source data in these cases led to the correct diagnosis. Artifacts caused by metallic prostheses are also minimized by the short echo time because the effect of signal loss due to dephasing is reduced (13).
Time-of-flight MR techniques enable successful visualization of the iliac and femoral veins but are susceptible to in-plane saturation and flow artifacts, particularly at the origins of the IVC and the common iliac veins (4,6). In a recent review of the diagnostic accuracy of two-dimensional time-of-flight MR imaging (7), an overall sensitivity of 97% and specificity of 93% for above-the-knee DVT were reported. However, to retain high in-plane spatial resolution and keep imaging times at acceptable levels, the protocols used in several of the studies involved gaps between sections (1–4). This means that 3D reconstruction techniques could not be used. Imaging time in the studies in which contiguous sections were used was increased to 30–45 minutes (5,6). In contrast, an imaging time of 4 minutes was used with VESPA MR venography in this study. MIP reconstruction of VESPA MR images is also enhanced owing to subtraction of the background signal.
An acute thrombus can occasionally appear very bright on T1-weighted MR images and may not be apparent after contrast material injection. The bright thrombus would, however, be clearly visible on the measurements acquired before contrast enhancement, and this problem was not encountered in our study.
Contrast-enhanced spiral computed tomographic (CT) venograms are also acquired in a series of two-dimensional sections. To reduce x-ray exposure, two-dimensional sections are frequently separated by gaps, and repeat acquisitions necessary for subtraction are not performed. Arterial signal therefore is not removed, and there is less scope for 3D reconstruction (14). Spiral CT and MR imaging after bilateral injection of contrast material into the pedal veins have also been reported (15,16). While high accuracy for the diagnosis of below- and above-the-knee DVT has been achieved with these techniques, they are invasive and share many of the problems of conventional ascending venography (16).
Several authors have found that the proximal extent of thrombosis may be underestimated at conventional venography and ultrasonography (US) compared with that depicted at MR and spiral CT venography (1,4,14,16–18). US images of the pelvic veins are frequently obscured by overlying structures, and, owing to a combination of dilution and distal obstruction, conventional venography frequently fails to demonstrate the upper extent of thrombosis. Use of bilateral venography, additional contrast material, and postural techniques may improve proximal opacification; however, these techniques require specialized skills and are not commonly used (4). Proximal vessels can be visualized despite distal obstruction by using VESPA MR venography because contrast material injected into a peripheral vein is distributed throughout the vascular system. VESPA MR venograms are also of high quality in the iliac veins and IVC because the large size of these vessels enables large volumes of contrast material to accumulate and the fixed position of these vessels in the pelvis minimizes the likelihood of movement artifacts and subtraction errors. In addition, with this technique, satisfactory images of the superficial veins and deep femoral and internal iliac veins are obtained bilaterally, allowing simultaneous visualization of all the above-knee veins.
Advantages of the VESPA MR Technique
The production of MIP venograms allows bilateral visualization of the entire femoral and iliac venous systems simultaneously, so reporting time is reduced. Images can be viewed at any angle, and the courses of small and tortuous vessels such as collateral vessels can be visualized. In addition, comparison between studies to monitor thrombus progression and resolution is facilitated. The high cost and low availability of MR imaging compared with those of other noninvasive diagnostic modalities limit its use. However, the use of a standard angiography sequence and standard software with this technique means that it could be used with the majority of clinical MR imagers with angiographic capabilities. Cost is also minimized by the very short imaging time.
VESPA MR venography is well suited for noninvasive assessment of the upper extent of above-the-knee thrombosis, which is frequently poorly visualized with conventional techniques. The upper extent of thrombosis would be important to assess in cases of possible thrombus progression and in other high-risk cases. Similarly, in patients who are being considered for thrombolysis or IVC filter placement, this technique could be used to assess the upper extent of thrombosis before intervention. Also, we have found that VESPA MR venography performed after intervention with catheter-directed thrombolysis and stent placement enables successful assessment of the patency of metallic stents. Artifacts caused by the metal stent are minimized by the short echo time (13).
We believe the 3D venograms produced with the VESPA MR venographic technique make this a very good technique for the visualization of complex venous anatomy in cases of chronic thrombosis and venous anomalies. VESPA MR venography can also be used as the sole evaluation for the exclusion of above-the-knee thrombosis if conventional tests are unsuccessful (eg, in patients who have full-length plaster leg casts or grossly swollen legs). In addition, early phase VESPA MR images yield high quality arteriograms, and background and enhanced images may reveal associated underlying disease or abnormality.
Future Applications
VESPA MR venography can be combined with other MR imaging techniques such as direct thrombus imaging and MR pulmonary embolus imaging to provide a comprehensive assessment of the chest and lower limbs in patients with venous thromboembolism (18,19).
Contrast enhancement of the vessel wall in cases of acute thrombosis but not in cases of chronic thrombosis was observed in all cases in this study and has been noted in previous studies (20). This observation may mean that VESPA MR venography and other contrast-enhanced techniques can enable the physician to distinguish acute thrombus from chronic thrombus in cases of recurrent thrombosis. Conventional examinations are frequently nondiagnostic in these cases.
In conclusion, contrast-enhanced MR venography with the VESPA subtraction technique enabled accurate diagnosis of femoral and iliac thrombosis and allowed studies to be interpreted on the basis of findings on MIP venograms—the sole images provided to the reviewers—in 95% of cases. This rapid 3D MR imaging technique is an important advance over time-of-flight MR imaging, which is usually performed as a series of two-dimensional section acquisitions separated by gaps. VESPA MR venography involves the use of standard angiography sequences and postprocessing software, does not require timing studies, and could be used with the majority of clinical MR imaging units with angiographic capabilities.
|
FOOTNOTES |
|---|
Abbreviations: IVC = inferior vena cava, DVT = deep venous thrombosis, MIP = maximum intensity projection, SFV = superficial femoral vein, 3D = three-dimensional, VESPA = venous enhanced subtracted peak arterial
Author contributions: Guarantors of integrity of entire study, D.G.W.F., A.R.M.; study concepts, D.G.W.F., A.R.M., A.L.M.; study design, D.G.W.F., A.R.M.; literature research, D.G.W.F.; clinical studies, D.G.W.F., A.R.M.; experimental studies, P.S.M., A.R.M.; data acquisition, D.G.W.F., A.R.M., P.S.M.; data analysis/interpretation, all authors; statistical analysis, D.G.W.F.; manuscript preparation, D.G.W.F.; manuscript definition of intellectual content, D.G.W.F., A.R.M., I.R.D.; manuscript editing, D.G.W.F., A.R.M.; manuscript revision/review and final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Remy-Jardin, M. Pistolesi, L. R. Goodman, W. B. Gefter, A. Gottschalk, J. R. Mayo, and H. D. Sostman Management of Suspected Acute Pulmonary Embolism in the Era of CT Angiography: A Statement from the Fleischner Society Radiology, November 1, 2007; 245(2): 315 - 329. [Full Text] [PDF]
|
|
|
|
 |
|
 A. Kluge, W. Luboldt, and G. Bachmann Acute pulmonary embolism to the subsegmental level: diagnostic accuracy of three MRI techniques compared with 16-MDCT. Am. J. Roentgenol., July 1, 2006; 187(1): W7 - 14. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Kluge, C. Mueller, J. Strunk, U. Lange, and G. Bachmann Experience in 207 combined MRI examinations for acute pulmonary embolism and deep vein thrombosis. Am. J. Roentgenol., June 1, 2006; 186(6): 1686 - 1696. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Pedrosa, M. Morrin, L. Oleaga, J. Baptista, and N. M. Rofsky Is True FISP Imaging Reliable in the Evaluation of Venous Thrombosis? Am. J. Roentgenol., December 1, 2005; 185(6): 1632 - 1640. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. P. Sheehy, G. E. Boyle, and J. F. M. Meaney Normal Anterior Spinal Arteries within the Cervical Region: High-Spatial-Resolution Contrast-enhanced Three-dimensional MR Angiography Radiology, August 1, 2005; 236(2): 637 - 641. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Obernosterer, M. Aschauer, H. Portugaller, H. Koppel, and R. W. Lipp Three-Dimensional Gadolinium-Enhanced Magnetic Resonance Angiography Used as a "One-stop Shop" Imaging Procedure for Venous Thromboembolism: A Pilot Study Angiology, July 1, 2005; 56(4): 423 - 430. [Abstract] [PDF]
|
|
|
|
|
|
M. R. Prince and H. D. Sostman MR Venography: Unsung and Underutilized Radiology, March 1, 2003; 226(3): 630 - 632. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
= Twelve of the patients with patent iliac and femoral veins had thrombosis isolated to the deep calf and/or popliteal veins.
