Applied Radiology

Dr. Glockner is an Assistant Professor and Consultant, and Dr. Lee is an Assistant Professor and Consultant, Department of Radiology, Mayo Clinic, Rochester, MN.

Magnetic resonance venography (MRV) is an often overlooked and underappreciated technique. While MR angiography (MRA) has generated enormous interest, almost since its inception, academic and clinical applications of MRV are relatively meager by comparison. This is unfortunate, since MRV can be highly accurate, easy to perform and successful in many situations where other imaging techniques yield ambiguous results.

Several methods are available to image veins with MRI, including contrast-enhanced (CE) and non–contrast-enhanced (NCE) pulse sequences, many of which can be employed effectively with little or no modification from existing body MRI or MRA protocols. The range of techniques available provides great flexibility, but can also be confusing for inexperienced users, and choosing the best pulse sequence for a given clinical situation can sometimes be difficult. This article gives brief descriptions of the most common techniques employed in MRV along with their strengths and weaknesses, with the goal of providing a framework for effective MRV in the clinical setting.

NCE techniques

When used without contrast, MRV can be divided into bright-blood and dark-blood techniques, with bright-blood methods generally more common and more reliable. Bright-blood NCE MRV pulse sequences include time-of-flight (TOF) gradient recalled-echo(GRE) or spoiled gradient recalled (SPGR) sequences as well as steady state free precession (SSFP) sequences.

Time-of-flight

Two-dimensional TOF represents one of the earliest MRV techniques, relying on inflowing blood to provide vascular signal.^1-3 TOF pulse sequences are SPGR or GRE acquisitions performed sequentially, i.e. all phase-encode steps are played out in a single slice before moving on to the next slice.This is less time efficient than multislice excitation, but results in much greater suppression of stationary tissue. Blood flowing into the slice is not saturated and appears bright relative to the dark (suppressed) background. For venography applications, a saturation band is applied above the slice to suppress inflowing arterial signal. TR (repetition time) can be adjusted to optimize venous contrast: Ideally, the volume of blood within the slice is replaced during the time between the initial radio frequency (RF)pulse and signal acquisition. For typical venous velocities, TRs in the range of 10 ms to 40 ms generally provide acceptable image quality. Longer TRs are more sensitive to slow flow, but also have less background suppression and longer acquisition times.

TOF venography has the advantage of simplicity. No special pulse sequences are required, and this technique is available on nearly every MR system. Thin sections (2 mm to 3 mm) can be acquired in the pelvis and extremities, allowing for 3-dimensional reconstructions without arterial contamination (Figure 1). Acquisition times are quite long if large volumes are covered, however, and this is a significant limitation in the chest and abdomen, where motion artifact is problematic. An alternative method is to prescribe thicker contiguous slices (6 mm to 8 mm) acquired during suspended respiration. Three-dimensional reconstructions are not possible in this case; however, the images are generally more than adequate to assess the patency of the inferior vena cava (IVC), hepatic veins and mesenteric veins (Figure 2).

Additional limitations of the TOF techniques include relative insensitivity to in-plane flow. The optimal acquisition plane lies orthogonal to the direction of flow, which is inefficient from the standpoint of acquisition time, and not always achievable.

Steady state free precession

Steady state free precession (SSFP) pulse sequences maintain the steady state of both longitudinal and transverse magnetization by application of a series of balanced RF pulses. These sequences have advantages over standard GRE or SPGR sequences in signal-to-noise ratio (SNR) and vascular contrast. SSFP image contrast is largely independent of TR and proportional to T2/T1. Since blood has a long T2 and a short T1, relative to most parenchymal tissue, vessels appear bright on SSFP images, with excellent SNR. Because the bright-blood effect of SSFP pulse sequences is primarily due to intrinsic relaxation properties of blood, inflow effects are less important, and artifacts related to in-plane and disordered flow are less prominent.^4,5

SSFP sequences require high-performance gradients to obtain very short TRs and TEs. As TR increases, image artifacts, including banding and susceptibility artifacts, become increasingly prominent, and acquisition times also increase. A second limitation of SSFP techniques is relatively poor background suppression. Arteries and fluid, in particular, also have high signal intensity, limiting the ability to generate 3-dimensional reconstructions. An additional limitation is occasional difficulty in detecting subacute thrombus, which can occur when the signal intensity of thrombus approximates that of blood (usually on the basis of shortened T1).

In our practice,we use both 2-dimensional and 3-dimensional SSFP sequences for assessment of veins in the chest, abdomen, and pelvis (Figure 3). Two-dimensional acquisitions are typically performed with 4 mm to 5 mm sections overlapping by 1 mm to 2 mm. Fat saturation is employed to reduce background signal from fat. These are rapid acquisitions, ideally performed during suspended respiration; however, image quality is usually acceptable in patients unable to hold their breath, particularly if parallel imaging is used. Three-dimensional SSFP sequences can be performed with thinner section thickness, 1 mm to 2 mm, allowing for near isotropic voxel acquisition and reformatting in additional planes. Acquisition times are longer, however, and fat saturation more difficult.

Recently, several vendors have introduced modified SSFP pulse sequences which employ inversion recovery pulses to suppress background tissue in the imaging volume as well as signal from inflowing venous blood.⁶ While initially intended for NCE MRA, these pulse sequences can be modified to acquire venous information by suppressing arterial rather than venous signal. Similarly, other techniques employing 3-dimensional FSE sequences with ECG gating to generate systolic and diastolic images can be modified to obtain venous information. The ability to generate 3-dimensional reconstruction of venous anatomy without contrast and without arterial contamination is very appealing, however, the efficacy of these techniques in body MRV has only begun to be evaluated.^7,8

Phase contrast pulse sequences are not commonly employed in MRV of the body, but are occasionally useful for quantification of velocity gradients across a stenosis and subsequent estimation of the pressure gradient.

Black blood techniques

Black blood images of veins are often obtained on routine FSE T2-weighted acquisitions. The physical basis of the black blood effect has to do with replacement of blood in the imaging slice during the echo time (TE). The black blood effect is complicated by the long echo trains used in some sequences as well as the phasicity of venous blood flow. These sequences were not designed to image vessels and, consequently, artifacts are relatively common, particularly in regions of slow flow.

ECG-gated double inversion recovery fast spin echo sequences are used in cardiac imaging to obtain black blood images.These employ a non-selective inversion pulse followed by a slice-selective pulse and fast spin echo readout after an inversion time TI. The inversion time is chosen to null the signal of blood, so that blood flowing into the imaging slice during TI appears dark. These techniques also work reasonably well for venous imaging and generally achieve a more consistent black blood effect than fast spin echo sequences (Figure 4).

Contrast-enhanced techniques

CE MRV is probably the most widely used technique, and is essentially identical to 3-dimensional CE MRA, employing a 3-dimensional spoiled gradient echo sequence in conjunction with a bolus of gadolinium-based contrast.^9-12 Vascular contrast results from the T1-shortening effects of gadolinium on adjacent water protons, and has relatively little dependence on inflow effects.

The simplest 3-dimensional CE MRV techniques involve 1 or more additional acquisitions after performing MRA. The contrast bolus is injected and MRA is performed when the concentration of gadolinium peaks in the arteries. Additional phases are then acquired until venous contrast is maximal (Figure 5). Alternatively, a test bolus or fluoroscopic triggering can be used to optimize venous rather than arterial signal. This reduces the total number of acquisitions but does not allow subtraction of arterial phase data. As with 3-dimensional CE MRA, acquisition times are generally short enough to be acquired during breath holding, which reduces or eliminates motion artifact. Since the vascular signal is not dependent on inflow effects, any acquisition plane can be chosen, thereby maximizing the efficiency of the scan. A limitation of CE MRV with respect to MRA is that the contrast bolus is more dilute by the time it reaches the venous system, and therefore the maximal contrast enhancement achieved in veins is typically lower than arteries. The addition of fat saturation to the 3-dimensional SPGR sequence can help to improve background suppression and emphasize vascular signal, at a small cost in acquisition time.

Three-dimensional reconstruction of CE MRV data is somewhat less straightforward than MRA reconstructions since the vein/background contrast is lower and there is usually arterial as well as venous enhancement. Reformatting and sub-volume maximum intensity projection images often are adequate to demonstrate venous anatomy and pathology. If pure arterial phase images are acquired, these can be subtracted from the optimal venous phase to generate a purely venous data set. Subtraction techniques rely on the assumption that there is no shift in position between the two acquisitions: when this is not the case (in poor breath holders, for example) substantial artifacts often result.

Direct MRV is a technique advocated by several authors in which a dilute bolus of gadolinium contrast (typically a 1:20 dilution in saline) is injected directly into a vein while continuously scanning the volume of interest.^13–14 This avoids the problem of contrast dilution that occurs when the contrast bolus first passes through the arterial system (Figure 6). The two major limitations of this technique are that venous access needs to be established in a peripheral vein of interest (typically the hand or foot), and that unless both arms orlegs are injected simultaneously there will be poor visualization of contralateral veins.

The techniques discussed above are summarized in Table 1.

Clinical considerations

Thoracic central veins are largely inaccessible to sonography, and MRV is an excellent technique for assessment of axillary, jugular,and subclavian veins, SVC, and pulmonary veins^10-11,15(Figures 6-7). ECG-gating, whether of SSFP or CE SPGR sequences, is often useful in reducing or eliminating artifact from cardiac motion. ECG-gated techniques are also useful for identifying intracardiac thrombus extending from below in patients with hepatic, adrenal, or renal tumors (Figure 4).

Assessment of portal and hepatic veins is usually accomplished within a standard hepatic protocol that includes 3-phase dynamic CE3-dimensional SPGR imaging (Figure 8). Our hepatic protocol also includes 2-dimensional fat-suppressed SSFP images, which provide an additional view of the portal and hepatic veins. Cine phase contrast sequences can demonstrate the direction of flow in the portal vein.

The IVC and renal veins are most commonly assessed in our practice in the setting of renal cell carcinoma staging.^9,16-17 We perform dynamic 3-dimensional fat-saturated SPGR sequences in the coronal plane (Figure 9). When significant IVC thrombus is present, it maybe necessary to acquire one or more additional delayed 3-dimensional SPGR acquisitions to visualize contrast enhancement of the iliac veins and proximal IVC, as venous return can be delayed. In our hands, 2-dimensional and 3-dimensional SSFP sequences are as effective at identification of IVC thrombus as CE 3-dimensional SGPR images (Figure 3), although the distinction between bland thrombus and tumor thrombus is sometimes more difficult.

Initial evaluation of the iliac and lower extremity veins is almost always performed with duplex sonography, which is usually sensitive and accurate in detection of venous thrombus, but can be limited in evaluation of pelvic and calf veins, obese patients, and chronic asymptomatic thrombus. Several studies have demonstrated the efficacy of MRV in these territories.^1-2,4,8,18 Common and effective techniques include 2-dimensional TOF MRV, 2-dimensional and 3-dimensional SSFP, and CE 2-dimensional or 3-dimensional SPGR pulse sequences (Figures 1, 10).

Comparison with other modalities

Sonography remains the test of choice for initial imaging of veins: it is accurate, readily available, and considerably less expensive than MRV. Visualization of central veins, particularly in the thorax, is often limited, however, and large patients may be particularly problematic.

CT venography has advantages of speed and spatial resolution in comparison with MRV. Many patients who are not candidates for MRV by virtue of pacemakers or other MRI incompatible devices, or claustrophobia, can be examined with CT venography. On the other hand, venous contrast-to-noise ratios (CNR) are almost always higher with MRV. Additionally, CT venography always requires the use of intravenous contrast, while many non-contrast methods are available with MRV, making MRV the preferred technique inpatients with renal insufficiency or contrast allergy. CT venography may also require 2 or more acquisitions to adequately capture contrast opacification of veins, thereby increasing the radiation dose.

Conclusion

MR venography is a valuable technique for assessing veins in regions inaccessible or poorly accessible to sonography. Both CE and NCE methods are widely available and can be performed quickly with high accuracy (Table 1). While the variety of pulse sequences available for MRV can be confusing, most should be familiar to radiologists performing body MRI, and they provide a flexibility unmatched by any other non-invasive technique.

REFERENCES

1. Carpenter JP, Holland GA, Baum RA, et al. Magnetic resonance venography for detection of deep venous thrombosis: comparison with contrast venography and duplex Doppler ultrasonography. J Vasc Surg. 1993;18:734-741.
2. Evans AJ, Sostman HD, Knelson MH, et al. Detection of deep venous thrombosis: prospective comparison of MR imaging with contrast venography. AJR. 1993;161:131-139.
3. Vogt FM, Herborn CU, Goyen M. MR venography. Magn Reson Imaging Clin N Am. 2005; 13:113-129.
4. Cantwell CP, Cradock A, Bruzzi J, et al. MR venography with true fast imaging with steady-state precession for suspected lower-limb deep vein thrombosis. J Vasc Interv Radiol. 2006; 17:1763-1769.
5. Wilson MW, LaBerge JM, Kerlan RK, et al. MR portal venography: Preliminary results of fast acquisition without contrast material or breath holding. Acad Radiol. 2002;9:1179-1184.
6. Miyazaki M, Lee VS. Nonenhanced MR angiography. Radiology. 2008;248:20-43.
7. Ito K, Koike S, Shimizu A, et al. Portal venous system: evaluation with unenhanced MR angiography with a single-breath-hold ECG-synchronized 3D half-Fourier fast spin-echo sequence. AJR. 2008;191:550-554.
8. Ono A, Murase K, Taniguchi T, et al. Deep vein thrombosis using noncontrast-enhanced MR venography with electrocardiographically gated three-dimensional half-fourierFSE: Preliminary experience. Magn Reson Med. 2009;e-published ahead of print.
9. Choyke PL, Walther MCM, Wagner JR, et al. Renal cancer: Preoperative evaluation with dual-phase, three-dimensional MR angiography. Radiology. 1997;205:767-771.
10. Shinde TS, Lee VS, Rofsky NM, et al. Three-dimensional gadolinium-enhanced MR venographic evaluation of patency of central veins in the thorax: Initial experience. Radiology. 1999;213:555-560.11.Lin J, Zhou KR, Chen ZW, et al. Vena cava 3D contrast-enhanced MR venography: A pictorial review. Cardiovasc Intervent Radiol. 2005;28: 795-805.
11. Ernst O, Asnar RV, Sergent G, et al. Comparing contrast-enhanced breath-hold MR angiography and conventional angiography in the evaluation of the mesenteric circulation. AJR. 2000;174:433-439.
12. Ruehm SG, Zimny K, Debatin JF. Direct contrast-enhanced 3D MR venography. Eur Radiol. 2001;11:102-112.
13. Tanju S, Sancak T, Dusunceli E, et al. Direct contrast-enhanced 3D MR venography evaluation of upper extremity deep venous system. Diagn Interv Radiol. 2006;12:74-
14. Valsangiacomo ER, Levasseur S, McCrindle BW, et al. Contrast-enhanced MR angiography of pulmonary venous abnormalities in children. Pediatr Radiol. 2003;33:92-98.
15. Laissy JP, Menegazzo D, Debray MP, et al. Renal carcinoma: Diagnosis of venous invasion with Gd-enhanced MR venography. Eur Radiol. 2000;10:1138-1143.
16. Hallscheidt PJ, Fink C, Haferkamp A, et al. Preoperative staging of renal cell carcinoma with inferior vena cava thrombus using multidetector CT and MRI: Prospectivestudy with histopathological correlation. J Comput Assist Tomogr. 2005; 29:64-68.
17. Kluge A, Mueller C, Strunk J, et al. Experience in 207 combined MRI examinations for acute pulmonary embolism and deep vein thrombosis. AJR. 2006;186:1686-1696.