Eur Radiol (2009) 19: 110–120 DOI 10.1007/s00330-008-1113-8
MAGN ETIC RE SONA NCE
Kjell-Inge Gjesdal Tryggve Storaas Jonn-Terje Geitung
A noncontrast-enhanced pulse sequence optimized to visualize human peripheral vessels
Received: 29 September 2007 Revised: 13 June 2008 Accepted: 22 June 2008 Published online: 15 August 2008 # European Society of Radiology 2008
J.-T. Geitung Department of Radiology, Haraldsplass University Hospital, Bergen, Norway
K.-I. Gjesdal (*) Sunnmøre MR-klinikk, N-6010 Aalesund, Norway e-mail:
[email protected] Tel.: +47-70-153690 Fax: +47-70-153691 T. Storaas Section for Diagnostic Physics, Department of Radiology, Ullevaal University Hospital, Oslo, Norway
Abstract The purpose of this paper is to present a pulse sequence optimized to visualize human peripheral vessels. The optimized MR technique is a 3D multi-shot balanced non-SSFP gradient echo pulse sequence with fat suppression. Several imaging parameters were adjusted to find the best compromise between the contrast of vascular structures and muscle, fat, and bone. Most of the optimization was performed in the knee and calf regions using multi-channel SENSE coils. To verify potential clinical use, images of both healthy volunteers and volunteers with varicose veins were
Introduction Magnetic resonance (MR) offers more ways to visualize human vessels than any other imaging technique. The early observation that blood flowing into the imaging slice produced less saturation of the MR signal than signals from stationary tissue gave rise to the time-of-flight method (TOF) [1]. The phase shift generated between spins moving along a magnetic gradient and non-moving spins gave us the phase contrast method (PCA), a method that can visualize both the blood vessels and their flow velocity [2]. The black-blood imaging method is a less-used MR angiographic (MRA) technique that exploits the loss of signal in blood vessels under certain parameter settings of a standard spin echo sequence [3].
produced. The balanced non-SSFP sequence can produce highspatial-resolution images of the human peripheral vessels without the need for an intravenous contrast agent. Both arteries and veins are displayed along with other body fluids. Due to the high spatial resolution of the axial plane source or reconstructed images, the need for procedures to separate arteries from veins is limited. We demonstrate that high signals from synovial joint fluid and cystic structures can be suppressed by applying an inversion prepulse but at the expense of reduced image signalto-noise and overall image quality. Keywords MR imaging . Vessels . Pulse sequence . MR angiography . MR venography
A bolus of gadolinium (Gd)-based contrast agents can reduce the T1 of blood from ∼1,200 ms at 1.5 Tesla to less than 1/10 of its original value depending on the dose [4]. If the bolus is imaged with a T1-weighted sequence at very short repetition time (TR) this effectively saturates all other signals except those from the bolus. The contrastenhanced MRA (CE MRA) is now widely accepted as the gold-standard method to image peripheral vascular disease [5–7]. Recently the first blood pool agent was commercially introduced [8]. Not only does this agent make first-pass, contrast-enhanced MRA possible, it also prolongs the imaging time window from a few seconds to nearly an hour, making high resolution images of both arteries and veins obtainable.
111
The new reports linking Gd-based contrast agents to nephrogenic systemic fibrosis (NSF) and the FDA’s black labeling of all such contrast agents have spurred renewed interest in developing noncontrast-enhanced high-spatial-resolution MRA techniques [9, 10]. The balanced steady-state free precession (b-SSFP) sequence, under the commercial names true FISP [fast imaging with steady precession (Siemens)], balanced FFE [fast field echo (Philips)] and FIESTA [fast imaging employing steady state acquisition (General Electric)], has received special attention from several investigators [11, 12]. It was soon recognized that the zero net gradient area in all three directions led to greatly reduced sensitivity to phase errors of constant moving spins and even turbulent flow [13]. Several studies have proved the technique’s ability to display the large vessels of the body [14–16], and balanced SSFP is so far the technique of choice to visualize the coronary arteries [17, 18]. Recently researchers have looked into the possibility of using balanced SSFP as a sequence for peripheral angiography, with a special focus on applying the technique on peripheral MR venography (MRV) [19–21]. It has been recognized that the sequence provides good bloodto-muscle contrast which, in combination with fat suppression, would make noncontrast-enhanced angiography feasible. Clinically peripheral MRV is of special interest with regard to studying the extent of deep venous thrombosis (DVT) and, preoperatively, chart-peripheral veins. MRV has the potential to detect perforators between deep and superficial veins. With a phase-contrast technique added to this, it is also possible to diagnose insufficient veins or valves. The need for a technique capable of imaging DVTs provided the motivation for developing the sequence presented in this paper.
The balanced non-SSFP sequence (b-TFE) A fat-suppressed multi-shot non-steady-state free precession technique with balanced gradients was optimized with the aim to visualize the human peripheral vessels. The vendor’s name for the sequence is balanced transient field echo (b-TFE). A graphic presentation of the b-TFE pulse sequence is seen in Fig. 1. The b-TFE sequence differs from the standard balanced SSFP (known commercially as b-FFE, True FISP, or FIESTA) sequence not only in the fundamental way that sampling of the echo takes place during a non-steady condition but also because of the wide choice of sequence parameters. This variety of image parameters gives the user an opportunity to be very precise in “fine tuning” the sequence to his or her specific imaging objective. Pulse sequence parameters vs. imaging objective The visualization of the peripheral vessels depends on several b-TFE sequence parameters. The use of high flip angles (>60°) to get a strong signal from the vessel lumen is well known in the clinic. Equally well known is the reduced sensitivity of the balanced sequences to phase errors due to flowing blood. However, to obtain a strong reduction in the signal from muscle, fat, free fluid, and bone calls for careful adjustment of several imaging parameters. These include the time of inversion, the number of start-up cycles, the shot duration, the shot delay and also which fat-suppression scheme is used. To study the effects of varying these parameters, both qualitative and quantitative methods were applied. Pulse sequence objectives
Material and methods
B-TFE sequence
Startup cycles
SPAIR fat suppression
B-TFE sequence
Delay
Optional inversion pulse
Images from three volunteers are presented in this study. The volunteers were two healthy men aged 24 and 48 years and one woman aged 44 years with varicose veins. Informed consent was obtained from the volunteers.
Shot duration
Startup cycles
Volunteers
Shot interval
SPAIR fat suppression
Imaging was performed on a 1.5-T whole-body MR system (Philips Intera, software release 2, Philips, Best, The Netherlands) fitted with maximum strength gradients of 33 mT/m and a maximum slew rate of 160 T m−1 s−1 (Nova gradients). For large imaging volumes, a four-channel SENSE-capable body array coil was applied. To allow high-spatial-resolution imaging, an eight-channel knee coil was selected.
Ideally an optimized pulse sequence for MRA purposes should be able to visualize not only the blood itself but also the vessel wall and the flow velocity of the blood.
Optional inversion pulse
Imaging hardware and software
Fig. 1 Graphic presentation of the 3D balanced TFE sequence. The SPAIR pulse is a fat-selected adiabatic inversion pulse
112
Table 1 The optimized 3D balanced non-SSFP parameters Parameter
Value
TR TE Flip angle SPAIR inv. delay Start-up cycles TFE factor Shot length Delay time Shot interval k-space sampling Half-fourier? Image resolution
Shortest Shortest 60-90° 160 ms 25 25–300 Less than 1,500 ms >1,500 ms >2,500 ms Low–high Yes Sub mm in-plane, 1–2 mm through-plane Radial 1,600 ms
Turbo direction Inversion prepulse delay
Further, one should easily be able to separate arteries from veins. No single sequence with all these features is available. Our study objective was to optimize the balanced non-SSFP such that the peripheral vessels were easily presentable as maximum intensity projections (MIPs) with a large volume of view and with a spatial resolution high enough to distinguish arteries from veins in the source images.
fat are utilized, and methods where the T1 of fat is used to null out or reduce the lipid signal. Recently a frequency adiabatic inversion prepulse [spectral attenuated inversion recovery (SPAIR)] was added to the MR system. This prepulse makes fat suppression independent of the B1 field and is less sensitive to coil quadrupole effects than, for instance, the SPIR sequence. When the fat signal is removed, the muscle appears relatively bright. Ideally one would choose magnetization transfer (MTC) prepulses to reduce the muscle signal but this technique in combination with balanced sequences is not a current option. The pulsatile motion of the arteries gives rise to slight blurring, especially when relatively long echo trains are applied. Therefore a compromise between sharpness and MR data acquisition time is needed as the latter increases quite dramatically with short acquisitions. Synovial fluid in the joints and cystic structures such as Baker cysts in the knee region have longer T1 and T2 values than blood and give rise to regions of strong signals in balanced gradient echo sequences. These signals can be “removed” by applying an inversion prepulse with the appropriate inversion time or with the use of postprocessing tools. The 3D balanced non-SSFP pulse sequence parameters Table 1 is a summary of the optimized parameters for imaging peripheral vessels using the balanced non-SSFP sequence.
Optimizing the balanced-TFE pulse sequence Balanced pulse sequences will, in general, produce a bright fat signal, relatively low signal from muscle, and high signal from most body fluids. Without suppression or postprocessing, the bright fat signal will reduce the usefulness of the angiographic MIPs. Fat suppression over large fields of view is not easy. This is especially true when imaging the lower extremities where profound variations in B0 due to anatomical factors are found and where achieving good B1 field homogeneity is not trivial. A modern MR system is equipped with several fat-suppression tools including water-selective prepulses, fat-suppression prepulses, binominal prepulses where differences in phase of water and
Results The effects of changing the imaging parameters are explained using both subjective and objective criteria. We used subjective criteria to evaluate the free-fluid removal techniques and to study the various fat-suppression schemes. The degree of fat-suppression is shown to be affected by the shot delay time. Visual inspection is sufficient to verify this effect. Measurements of the signal intensities of blood, muscle, fat, and bone were performed to produce numbers quantifying the contrast effects when changing the number
Table 2 Contrast values between blood and tissue for different numbers of start-up cycles Contrast values
Sblood − Smuscle/Sblood Sblood − Sbone/Sblood Sblood − Sfat/Sblood
Start-up cycles 0
Default
25
50
0.25 0.80 0.79
0.63 0.79 0.95
0.70 0.70 0.95
0.85 0.36 0.36
113
Fig. 2 Free fluid removal. a The unaltered image. b A 3D post-processing tool has been applied prior to making the maximum intensity projection. c The result following an inversion pulse with 1,600 ms inversion time before the balanced non-SSFP sequence
Fig. 3 Different fat-suppression schemes. a-c Two, three, and four binominal pulses, respectively, were used prior to the pulse sequence. d SPIR fat-suppression. e The SPAIR prepulse is used
114
Fig. 4 The effect of varying the number of start-up cycles. a No start-up cycles were applied resulting in a high muscle signal and blurring from short T1 components. b Fifty start-up cycles were used, resulting in loss of fat-suppression. c The default value has been used. d Twenty-five start-up cycles are repeated prior to signal sampling providing slightly better muscle-signal suppression than the default value
Fig. 5 The effect of varying the shot length. Increasing the shot length drastically reduces the overall scan time but also makes the pulse sequence more vulnerable to blurring and pulsatile effects as seen in these images
115
of start-up cycles (Table 2). Altering the parameter ‘shot length’ gave rise to a variable degree of vessel blurring, and this effect was measured and visualized using intensity profiles. At the end of the section, two large field-of-view MIP images are displayed. Further, a very high resolution image of the leg vessels is presented. These images demonstrate the power of this method to display human peripheral vessels without any need for an extravascular contrast agent. Free fluid removal Figure 2 presents the results of two very different methods for removal of free fluid (Fig. 2b, c) and compares these results to an unaltered image (Fig. 2a). The central image (Fig. 2b) is the result of manually removing the highintensity fluid areas from the 3D dataset. The time consumption was less than a minute for the example in Fig. 2 but depends very much on the post-processing tools that are available in the clinic. The balanced non-SSFP sequence itself can be altered such that discrimination between free fluid and blood is achievable. This is done by applying an inversion prepulse with an inversion time set to null out the signal from long T1 components. In the example in Fig. 2c, the inversion time was set to 1,600 ms.
Fig. 6 The effect of lengthening the shot delay. Good fat suppression depends on this parameter as seen in these images (arrow). A shot delay of more than 1,500 ms is used in our standard protocol
While the first method removes the free fluid at the expense of the radiologist’s time, the latter method also removes the water but with a less satisfactory fat-suppression result. Fat suppression Figure 3 presents MIPs of the balanced non-SSFP sequence using different lipid suppression strategies. Figure 3a-c demonstrates fat suppression with a variety of binominal prepulses, Fig. 3d using spectral saturation pulse (SPIR), and finally Fig. 3e showing the result after applying an adiabatic inversion pulse prior to signal detection (SPAIR). Homogeneous fat suppression is not achieved with binominal prepulses as can be observed in Fig. 3a-c. Both the SPIR and the SPAIR techniques offer essentially equally homogeneous fat suppression and are therefore preferred. Start-up cycles The number of start-up cycles prior to signal sampling is usually set to “default” and calculated by the MR system’s software to produce optimal image quality. Our motivation for adjusting this parameter was based on simulations that
116
indicated that the muscle signal could be reduced by increasing the number of cycles. Figure 4 presents results from different numbers of start-up cycles. Additionally, Table 2 shows numerical values for the contrast between blood and relevant tissues. Contrast-to-noise measurements were not possible be-
Fig. 7 A six-station MIP visualization stitched together (left) along with axial source images at four different levels (right)
cause parallel imaging was applied. As the number of start-up cycles increases, the desired contrast between blood and muscle also increases but at the expense of a fluctuating fat signal. The best compromise is approximately 25 start-up cycles, which is the number used in our standard protocol.
117
Shot length The shot length includes the fat-suppression part plus the number of repetitions of the balanced sequence including the start-up cycles. Figure 5 reveals the effect of changing the shot length while keeping all other parameters constant. As the shot length is increased, a more pronounced blurring is observed. The data for this study are presented as the intensity profile through central arteries and veins of the leg. As the blurring becomes more pronounced at longer shot lengths, the loss in vessel resolution can be easily observed in the shape of the intensity profile. For highest image quality, the shot length is kept to ∼1,100 ms. Shot delay time and shot interval The shot delay time in combination with shot length determines the shot interval. In Fig. 6, images produced at Fig. 8 Multi-station MIPs (left) and 3D segmentation (right) illustrating varicose veins
several delay times are presented.The most visible effect is the improved fat suppression (see arrow) as the shot delay is increased, along with an overall better SNR and angiographic image quality obviously at the expense of longer imaging time. Large field-of-view images Figure 7 displays the MIPs resulting from imaging the peripheral vessels of a 25-year-old healthy male volunteer from the pelvic plane down to the feet in six different stations. Examples of source images at the pelvis, thigh, knee, and leg level are shown in the same figure. Pixel size was 1.0 × 1.0 mm and slice thickness set at 1.5 mm, SENSE factor = 2, and 167 slices per station, resulting in a scan time of 2 min:33 s per station and a total imaging time of 15 min:18 s for all six stations.
118
Fig. 9 Very high-resolution images (voxel size=0.175 mm3) of the leg vessels. a A coronal slice with a corresponding axial slice (b). c MIP of a zoomed-in area of b angled for best discrimination between the fibular artery and its vein pair. The profile (in red) verifies the excellent spatial resolution of the source images
agents [4, 22]. After nearly 10 years of use, the method is regarded as both safe and diagnostically accurate [22]. However, the rapid passage of the T1-shortening contrast bolus results in a narrow imaging time window. This makes the method technically demanding and results in images with a relatively coarse imaging matrix, typically 5– Very high-resolution imaging 10 mm3 with 1.5- to 3-mm slice thickness even when As one attempts to achieve even higher spatial resolution, strong gradients and parallel imaging are used. Reformatthe TR of the pulse sequence is increased and thereby so ting the coronal source images into more easily readable too is the shot length. One should therefore slightly reduce transverse source images is, for this reason, of little the number of TRs in the echo train such that the length of radiological value. MRA has already demonstrated its the echo train is kept at approximately 1,500 ms or less. superiority to visualize peripheral arteries. It is, however, Figure 9 illustrates both graphically (see signal intensity always necessary to have as good a resolution as possible profile) and visually how accurately the fibular artery and before performing an angioplasty. With Doppler, CT vein pair can be resolved using the balanced non-SSFP angiograms, and MRA, it is possible to obtain excellent sequence. Using the parameters 300 slices, TR/TE= unenhanced imaging. What remains is to obtain even better 6.0 ms/3.0 ms, flip angle=80°, a shot length of 919 ms, spatial resolution of stenosis and, if possible, to combine delay time of 2,313 ms and a SENSE factor of 3 gave a data such anatomical data with flow measurements. acquisition time of 9 min:17 s. Voxel size was 0.175 mm3. The demonstration of both peripheral and central veins before performing surgery is more challenging than both demonstrating arteries and looking for DVT. Excellent anatomical mapping is needed as well as demonstration of Discussion possible insufficiencies in the central veins. MRV of the lower peripheral structures is of interest with MR imaging of the peripheral vessels is most frequently performed using contrast enhancement with intravenous regard to DVT and may replace conventional venograms. In Fig. 8, MIP multi-station images of a 44-year-old female volunteer with several varicose veins are shown together with a 3D segmented close up of such a vein.
119
Peripherally in the leg and ankle we know that ultrasound with Doppler is suboptimal and contrast venograms may still be needed. The most promising part is, however, the ability to do nearly perfect mapping of all central and peripheral veins in the lower extremities. Conventional intravenous contrast venography serves as the gold standard. However, the technique is invasive, partly operator dependent and not without risks, including the possibility of procedure-induced DVT [23]. Several MRI imaging techniques have been applied to the peripheral venous system including time-of-flight (TOF), phase contrast (PC), and fast spin echo sequences (FSE) [24]. These methods have demonstrated limited clinical value because of artifacts due to complex flow, long data acquisition times, limited spatial resolution, and suboptimal contrast sensitivity [25]. The year 2005 saw the introduction in Europe of the first commercially available intravascular contrast agent, referred to as the blood pool agent. Gadofosveset trisodium (Vasovist, Bayer Schering Pharma) binds to the albumin in the blood and by this prolongs the potential imaging window from seconds to nearly an hour [26]. The contrast agent can be used as an arterial-phase, first-pass bolus agent with improved T1 shortening. Equally interesting is the long imaging window during which the contrast is slowly cleared from the blood. This long period makes images of exquisite resolution possible but at the cost of displaying both arteries and veins. Previously, the venous signal was seen as ‘contamination’ but with the potential of imaging at sub-1 mm3 resolution, the radiologist can relatively easily separate arteries from veins in the axial (reformatted) plane just by inspecting the source images. Several articles were recently published as a supplement of European Radiology (Springer-Verlag Berlin Heidelberg) including one article dealing with the technical requirements, biophysical considerations, and protocol optimization for using gadofosveset trisodium and a second article focusing on imaging of the peripheral vessels [26, 27]. The latter article concluded that “the potential advantages of contrast-enhanced MRA in PAD (peripheral arterial disease) are further increase by the use of blood-pool agents as gadofosveset, which allow for both first-pass and steady-state imaging over large field of view”. The full clinical impact of gadofosveset trisodium has not been documented so far, but in general the high spatial resolution of these images opens up the possibility for a more detailed examination of vascular anatomy and pathology than previously possible. A small (less than 70% and as small as 10–20% of the lumen) stenosis may be reexamined with higher resolution or with such a high initial spatial resolution that we can describe a stenosis better than any previous method has done. To give a good map of all perforators between central and deep veins as well as to map all peripheral and muscle veins will be a valuable supplement to existing methods. If the latter can be
combined with PC-MRV, it may become the single method for preoperatively examining veins. However, the recent reports that Gd-based contrast media at approved doses may be nephrotoxic for patients with severe kidney failure is expected to create a greater interest in noncontrast-enhanced sequences [10, 11]. Balanced gradient echo sequences (B-FFE, True-FISP, FIESTA) have the potential to offer high-resolution images with a strong signal from blood and therefore are the most likely candidates to compete with contrast-enhanced MRA. Our initial results suggest that the balanced non-SSFP sequence produces images of higher spatial resolution than the CE-MRA method. The sequence generally has a higher signal-to-noise ratio (SNR) than both the inflow and the phase angiography images but does not discriminate between arteries and veins. Nor can the method in its present form be used to quantify flow. However, the highresolution source images along with MPR images give the radiologist a unique insight into morphological details of the peripheral vessels that has not previously been possible with other MR imaging methods. Another unique feature of the balanced non-SSFP sequence is its ability to image very slow or non-flowing blood. This is of special interest with regard to preoperative mapping of veins, as previously described. A large portion of the work behind this presentation of the balanced non-SSFP sequence has been related to studying the behavior of the fat signal. Fat suppression is of importance with regard to vessel conspicuity and to the quality of the MIP. Our experience is that the new, bettercontrolled B1–field multi-channel coils are essential for homogeneous fat suppression. The imaging results obtained for this article do not involve modification of the scanner in terms of pulse sequence development, rather only careful rendering of MR parameters of the balanced non-SSFP sequence available within the system. The images are of such quality that a comparison study of the technique to highresolution gadofosveset-trisodium-enhanced images is relevant. Such a study is now taking place.
Conclusion In conclusion, we have optimized a non-SSFP balanced gradient echo sequence for the purpose of obtaining highresolution noncontrast-enhanced images of the peripheral vessels. The sequence is currently being used in a DVT study comparing its clinical value to a blood-pool contrast agent. Potential conflict of interest The MR-clinic (where the first author of this paper is working) has a clinical science research agreement with Philips Medical Systems, Best, Holland.
120
References 1. Dumoulin CL, Cline HE, Souza SP, Wagle WA, Walker MF (1989) Threedimensional time-of-flight magnetic resonance angiography using spin saturation. Magn Reson Med 11(1):35–46 2. Dumoulin CL, Souza SP, Walker MF, Wagle W (1989) Three-dimensional phase contrast angiography. Magn Reson Med 9(1):139–149 3. Edelman RR, Mattle HP, Wallner B (1990) Extracranial carotid arteries: evaluation with “black-blood” MR angiography. Radiology 177:45–50 4. Prince MR (1994) Gadoliniumenhanced MR aortography. Radiology 191:155–64 5. Ruehm SG, Wiesner W, Debatin JF (2000) Pelvic and lower extremity veins: contrast-enhanced threedimensional MR venography with a dedicated vascular coil—initial experience. Radiology 215:421–427 6. Lebowitz JA, Rofsky NM, Krinsky GA, Weinreb JC (1997) Gadoliniumenhanced body MR venography with subtraction technique. AJR Am J Roentgenol 169:755–758 7. Fraser DGW, Moody AR, Davidson IR, Martel AL, Morgan PS (2003) Deep venous thrombosis: diagnosis by using venous enhanced subtracted peak arterial MR venography versus conventional venography. Radiology 226:812– 820 8. Nikolaou K, Kramer H, Grosse C, Clevert D, Dietrich O, Hartmann M, Chamberlin P, Assmann S, Reiser MF, Schoenberg SO (2006) High-spatialresolution multistation MR angiography with parallel imaging and blood pool contrast agent: initial experience. Radiology 241(3):861–72 2006 Dec 9. Thomsen HS (2004) Gadolinium-based contrast media may be nephrotoxic even at approved doses. Eur Radiol 14 (9):1654–1656
10. United States Food and Drug Administration(2007) Public health advisory. Gadolinium-containing contrast agents for magnetic resonance imaging (MRI): Omniscan, OptiMARK, Magnevist, ProHance and Multihance. http://www. fda.gov/cder/drug/advisory/gadoliniu m_agents.htm. Accessed 9 July 2008 11. Duerk JL, Lewin JS, Wendt M, Petersilge C (1998) Remember true FISP? A high SNR, near 1-second imaging method for T2-like contrast in interventional MRI at 2 T. J Magn Reson Imaging 8(1):203–208 12. Bieri O, Markl M, Scheffler K (2005) Analysis and compensation of eddy currents in balanced SSFP. Magn Reson Med 54:129–137 13. Haacke EM, Wielopolski PA, Tkach JA, Modic MT (1990) Steady-state free precession imaging in the presence of motion: application for improved visualization of the cerebrospinal fluid. Radiology 175:545–552 14. Barkhausen J, Quick HH, Lauenstein T, Goyen M, Ruehm SG, Laub G, Debatin JF, Ladd ME (2001) Whole-body MR imaging in 30 seconds with real-time true FISP and a continuously rolling table platform: feasibility study. Radiology 220(1):252–256 15. Pereles FS, McCarthy RM, Baskaran V, Carr JC, Kapoor V, Krupinski EA, Finn JP (2002) Thoracic aortic dissection and aneurysm: evaluation with nonenhanced true FISP MR angiography in less than 4 minutes. Radiology 223 (1):270–274 16. Scheffler K, Lehnhardt S (2003) Principles and applications of balanced SSFP techniques. Eur Radiol 13 (11):2409–2418 17. Deshpande VS, Shea SM, Laub G, Simonetti OP, Finn JP, Li D (2001) 3D magnetization-prepared true-FISP: a new technique for imaging coronary arteries. Magn Reson Med 46(3):494– 502 18. Zagrosek A, Noeske R, Abdel-Aty H, Friedrich MG, Dietz R, Schulz-Menger J (2005) MR coronary angiography using 3D-SSFP with and without contrast application. J Cardiovasc Magn Reson 7(5):809–814
19. Spuentrup E, Buecker A, Stuber M, Gunther RW (2001) MR-venography using high resolution true-FISP. Rofo Fortschr Geb Rontgenstr Nuklearmed 173(8):686–690 20. Gjesdal, KI, Storås T, Geitung JT (2003) A pulse sequence for MR venography. In: Proceedings of the 20th Annual Scientific Meeting of the ESMRMB, Rotterdam, 18–20 Sept 2003, p 160 21. Hargreaves BA, Vasanawala SS, Nayak KS, Hu BS, Nishimura DG (2003) Fatsuppressed steady-state free precession imaging using phase detection. Magn Reson Med 50(1):210–213 22. Ersoy H, Zhang H, Prince MR (2006) Peripheral MR angiography. J Cardiovasc Magn Reson 8:517–528 23. Redman HC (1988) Deep venous thrombosis: is contrast venography still the diagnostic “gold standard”? Radiology 168(1):277–278 24. Evans AJ, Sostman HD, Witty LA et al (1996) Detection of deep venous thrombosis: prospective comparison of MR imaging and sonography. J Magn Reson Imaging 1:44–51 25. Bluemke DA, Wolf RL, Tani I, Tachiki S, McVeigh ER, Zerhouni EA (1997) Extremity veins: evaluation with fastspin-echo MR venography. Radiology 204:562–565 26. Rohrer M, Geerts-Ossevoort L, Laub G (2007) Technical requirements, biophysical considerations and protocol optimization with magnetic resonance angiography using blood-pool agents. Eur Radiol Suppl 17:B7-B12 27. Giovagnoni A, Catalano C (2007) Application of blood-pool agents in visualization of peripheral vessels. Eur Radiol Suppl 17:B18-B23
