HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Schoenberg, S. O. Articles by van Kaick, G. Search for Related Content PubMed PubMed Citation Articles by Schoenberg, S. O. Articles by van Kaick, G. Related Collections Related Article

Renal Arteries: Optimization of Three-dimensional Gadolinium-enhanced MR Angiography with Bolus-timing–independent Fast Multiphase Acquisition in a Single Breath Hold¹

¹ From the Department of Radiology, German Cancer Research Center (dkfz), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany (S.O.S., M.B., M.V.K., M.E., H.H., I.Z., G.v.K.); Siemens Medical Systems, Erlangen, Germany (G.L.); and the Department of Surgery, University Hospitals, Heidelberg, Germany (F.K.). Received December 3, 1997; revision requested January 22; revision received September 21; accepted November 5. Supported in part by a grant from the Verein zur Foerderung der Krebserkennung und Krebsvorsorge, e.V. Address reprint requests to S.O.S.

	Abstract

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
PURPOSE: To compare two different three-dimensional (3D) gadolinium-enhanced magnetic resonance (MR) angiographic techniques.

MATERIALS AND METHODS: In 26 patients suspected of having renal artery stenosis, results with fast multiphase 3D MR angiography were compared to those with standard 3D MR angiography in 37 patients. With both techniques, 31-second breath-hold acquisitions were performed. Multiphase angiography comprised five discrete 6.4-second acquisitions without bolus timing, and standard angiography comprised a single acquisition based on test-bolus timing. Two readers evaluated images obtained with both techniques in terms of image quality, artifacts, and vessel conspicuity. Accuracy of findings on the multiphase 3D MR angiograms for assessment of renal artery stenosis was determined by comparing them to digital subtraction angiograms and surgical findings.

RESULTS: In the early arterial phase, multiphase 3D MR angiograms showed no image degradation by venous overlay, whereas standard 3D MR angiograms depicted at least minor overlay in 53 of 83 renal arteries (P < .001). Less parenchymal enhancement in the early arterial phase resulted in a higher vessel conspicuity for the divisions and segmental arteries (P < .001). Both readers detected and correctly graded 18 of 20 stenoses on the multiphase angiograms with almost perfect interobserver agreement ( > 0.89).

CONCLUSION: Renal multiphase 3D MR angiography is an accurate technique requiring no bolus timing. The performance of early arterial phase imaging leads to improved depiction, particularly of the distal renovascular tree, compared to that with standard single-phase 3D MR angiography.

Index terms: Magnetic resonance (MR), maximum intensity projection, 961.12949 • Magnetic resonance (MR), three-dimensional, 961.12942, 961.149, 961.721 • Magnetic resonance (MR), vascular studies, 961.12942, 961.149, 961.721 • Renal arteries, MR, 961.12942, 961.149, 961.721

	Introduction

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Magnetic resonance (MR) angiography is a noninvasive and nonionizing alternative modality for imaging of the renal arteries. Although remarkable results have been reported with standard time-of-flight and phase-contrast techniques for the detection of proximal renal artery stenosis (1–4), they have not been accepted in clinical routine as a substitute for digital subtraction angiography (DSA) for definitive presurgical evaluation. Depiction of smaller vessel details with the techniques was limited due to in-plane saturation and motion artifacts during long acquisition times.

With the introduction of three-dimensional (3D) gadolinium-enhanced MR angiography (5), these limitations were overcome because a contiguous 3D data set could be acquired in a single breath hold (6–8). The use of contrast media minimizes saturation effects even in distal vessels and thus allows depiction of small vessel structures. Timing of the contrast media arrival is crucial, since sufficient T1 shortening of blood is achieved only during peak vascular concentrations of standard gadolinium chelates. In addition, delayed acquisition leads to venous overlay, which substantially decreases the diagnostic accuracy. Different strategies have been developed for accurate bolus timing, including the test-bolus technique (9–11), automated detection of the contrast media (12), and real-time depiction of the bolus arrival (13). With these techniques, a diagnostic accuracy comparable to that of DSA has been reported for the detection of renal artery stenosis (10,11) and stent patency (14). However, due to acquisition times of 20–30 seconds, substantial parenchymal enhancement limits the delineation of distal and intrarenal arteries even if optimal bolus timing is achieved. Because of this, the diagnostic accuracy for stenosis detection in these locations has not been definitely evaluated. In addition, the general clinical applicability is limited by the complexity of these techniques.

To increase the temporal resolution, k-space interpolation schemes can be used that substantially decrease imaging time for a single 3D data set (15). With this technique, no bolus timing is necessary, since multiple data sets for different phases of contrast media passage can be obtained. This technique has required complex postprocessing steps and is subject to a variety of artifacts, since the peripheral lines of k space are sampled less frequently than is the k-space center.

Current high-performance gradient systems allow use of short repetition times that enable fast multiphasic acquisitions of 3D data sets without k-space interpolation. To acquire complete 3D data sets in only seconds, asymmetric data sampling in k space is employed. In this study, multiple renal artery 3D data sets of the different arterial, parenchymal, and venous phases were acquired in a single breath hold without prior bolus timing or off-line data postprocessing. The results are compared to those with a standard test-bolus 3D MR angiographic technique.

	MATERIALS AND METHODS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
All MR examinations were performed on a 1.5-T system (Magnetom Vision; Siemens Medical Systems, Iselin, NJ) equipped with a resonant echo-planar imaging gradient overdrive (maximum gradient strength, 25 mT/m; shortest rise time to maximum, 300 µsec). In all examinations, the standard four-element phased-array body coil was centered at the level of the renal arteries.

Renal MR Angiographic Studies
For depiction of the kidneys and the aorta, a series of 20 sagittal breath-hold, T1-weighted, fast low-angle shot (FLASH) images were obtained. Axial breath-hold, T2-weighted, turbo spin-echo images were then obtained at the level of the kidneys for further characterization of any identified renal lesion. On the basis of the sagittal localizer images, the 3D MR angiographic slab was prescribed in the coronal plane. Because the renal arteries generally arise proximal and always posterior to it, the ventral border of the distal aorta was used as a landmark for the anterior extension of the 3D imaging volume. As much of the kidneys as possible was covered in the defined imaging time. In two different groups of patients, one of the following 3D MR angiographic techniques was applied.

Standard 3D gadolinium-enhanced MR angiography with test bolus.—A previously described (11) standard 3D MR angiographic sequence, which is routinely available on this MR system (repetition time msec/echo time msec = 5/2; read-out bandwidth, 488 Hz per pixel), was set up with phase-encoding direction from left to right and 3D partition-encoding in the anteroposterior direction. To prevent extensive aliasing from the abdominal tissue outside the kidneys in the phase-encoding direction, a rectangular field of view of at least 24–27 x 32–36 cm was necessary. Aliasing from the patient's arms was suppressed by using copper mesh pillows. For maximum coverage of the vascular anatomy with adequate spatial resolution, a 3D slab thickness of 8–9 cm was used with 44–50 partitions, resulting in an effective partition thickness of 1.8 mm. With a 180 x 256 matrix (135 phase-encoding lines for a three-quarter rectangular field of view), a total imaging time of about 29–33 seconds and a nearly isotropic spatial resolution of 1.5 x 1.9 x 1.8 mm were achieved. Because not all patients can reliably hold their breath for the total imaging time, centric reordering of k space was implemented, since this acquisition scheme reduces artifacts from incomplete breath holds (16). As the contrast information was encoded in the central k-space lines, which are sampled first, the maximum gadolinium chelate concentration had to be timed with the beginning of the 3D MR angiographic acquisition.

The transit time of the contrast media was determined by using a test bolus. One milliliter of gadopentetate dimeglumine (Magnevist; Berlex Laboratories, Wayne, NJ) was injected in a cubital vein with an automated infusion system (CAI 626; Doltron, Uster, Switzerland) at a standardized infusion rate of 3 mL/sec. Simultaneously with the beginning of the infusion, imaging with an axial two-dimensional turbo FLASH sequence (8.5/4/100 [inversion time msec], field of view of 35 cm, section thickness of 8 mm) was performed at the level of the renal arteries, acquiring one image per second over 60 seconds. Arrival time of the contrast media at the level of the renal arteries was determined by using a dynamic region-of-interest analysis of the signal intensity in the aorta. Considering the length of the gadolinium chelate infusion and the duration of the bolus, the delay between the beginning of the infusion and the start of the 3D MR angiographic measurements was calculated (in accordance with findings in previous studies of centrically reordered sequences [12]) as the time from the start of the infusion to one-third of the maximum signal intensity peak in the aorta. This value was chosen to prevent artifacts as a result of rapidly changing gadolinium chelate concentrations in too-early acquisition of central k space and to not miss peak enhancement. Patients were asked to suspend breathing during the whole acquisition time.

Multiphase 3D gadolinium-enhanced MR angiography without test bolus.—The multiphase acquisition without bolus timing was performed with a 3D fast FLASH sequence by using an asymmetric k-space acquisition in the read-out, phase-encoding, and partition directions. In the read-out direction, only 160 data points were acquired with the gradient echo occurring at point 32, resulting in a repetition time of 3.2 msec and an echo time of 1.1 msec at a read-out bandwidth of 650 Hz per pixel. For reconstruction, raw data were filtered and zero filled asymmetrically to 256 points. This asymmetric sampling and reconstruction strategy was also used in the phase-encoding and partition-encoding directions, where approximately five-eighths of k space was acquired starting at about one-eighth before k-space center. Additionally, a partially self-refocusing, radio-frequency pulse was used for excitation of the 3D slab, which substantially shortened the echo time.

To define an imaging time short enough for optimum depiction of the proximal and distal renal arteries independent of bolus timing, the intravascular signal intensity versus time after injection of gadolinium chelate was analyzed at different locations in the renovascular tree. In five prior patients, not enrolled in this study, imaging was performed with a two-dimensional dynamic FLASH sequence (5/2, matrix of 200 x 256, flip angle of 40°) with 128 repetitions. One image was acquired per second with the beginning of an automated infusion of gadopentetate dimeglumine (0.2 mmol per kilogram of body weight) at an infusion rate of 3 mL/sec. From the time–signal intensity plot (Fig 1), a time window was chosen of 6 seconds for the acquisition of one 3D data set, since no enhancement of venous structures was observed during this acquisition period, while the arterial signal intensity was at maximum. With the 3D fast FLASH sequence, 90 phase-encoding lines and 22 partitions could be measured for the defined imaging time, resulting in a reconstructed voxel size after zero filling identical to that obtained with the standard 3D MR angiographic sequence. The other parameters, including field of view, position of the 3D slab, and slab thickness, were identical to those used in the standard sequence. In a multiphasic acquisition independent from bolus timing, four to five repetitive 6.4-second measurements were performed with an intermittent delay of 150 msec, leading to a total imaging time of 25–31 seconds. In all measurements, the multiphasic acquisition was started 8 seconds after the beginning of the contrast media infusion. The time delay had to be short enough to ensure appropriate detection of the arterial phase in patients with short circulation time and long enough to maximize the monitoring period in patients with delayed bolus passage. The dose and infusion rates were identical to those used in standard 3D MR angiography. Patients were asked to suspend breathing with the beginning of the measurement.

View larger version (22K): [in this window] [in a new window]   Figure 1. Graph depicts variation in signal intensity difference for various vessel regions as a function of time during first pass of the gadopentetate dimeglumine bolus (0.2 mmol/kg at an infusion rate of 3 mL/sec). The curves were low-pass filtered to remove high-frequency fluctuations. The dotted lines indicate, in descending order, the variation in signal intensity difference in the aorta, proximal and distal renal artery, and intrarenal segmental artery. The solid line indicates parenchymal enhancement. The two dashed lines represent, in descending order, a segmental renal vein and the main renal vein. The time delay between maximum arterial and venous enhancement is within a range of 6 seconds. Overlay of the renal parenchyma (light and dark gray box) already occurs 2–3 seconds after the onset of the pure arterial vascular phase, which significantly alters the further signal intensity curve in the distal and intrarenal arterial vessels.

Patients
All patients were referred from the Department of Surgery, Division of Vascular Surgery, University Hospitals, with known or suspected renal artery disease. Clinical suspicion for renal artery stenosis included hypertension, poorly controlled with medication or rapidly deteriorating, and progressive renal insufficiency. The study protocol was approved by the institutional review board, and informed consent was obtained.

Standard 3D gadolinium-enhanced MR angiography with test bolus.—From October 1996 through April 1997, a total of 37 patients (27 men, 10 women; age range, 21–74 years; mean age, 59 years ± 21 [SD]) were evaluated with the standard test-bolus 3D MR angiographic technique. Six patients (four men, two women) were evaluated before and after surgical revascularization of the renal arteries. Serum creatinine level in the 43 patients ranged from 0.6 to 3.8 mg/dL (mean, 1.4 mg/dL ± 0.8). Among the 37 preoperative and six postoperative cases, a test bolus was used in 41 cases to define the transit time of the contrast media prior to the standard 3D MR angiographic measurement. In two patients with mild claustrophobia, the determined preoperative transit time was also used as the postoperative measurement to minimize total imaging time. Mean imaging time for a single measurement was 27 seconds (range, 24–31 seconds). Since two of the postoperative patients underwent nephrectomy and one other patient initially presented after nephrectomy, 83 kidneys could be evaluated with a total of 96 renal arteries including 13 accessory vessels.

Multiphase 3D gadolinium-enhanced MR angiography without test bolus.—From May through October 1997, 26 patients (17 men, nine women) underwent renal 3D MR angiography with the multiphase method. Four of these patients (two men, two women; mean age, 55 years ± 16; age range, 34–67 years) were also examined before and after surgical revascularization. Their serum creatinine levels ranged from 0.8 to 3.4 mg/dL (mean, 1.5 mg/dL ± 1.1). A total of 30 examinations (26 preoperative, four postoperative) were performed. The mean imaging time for a single phase was 6.4 seconds (range, 5.5–6.8 seconds). With a mean of five measurement phases (range, four to six), the mean total imaging time was 31.5 seconds (range, 22–34 seconds). For all multiphase acquisitions, an imaging delay of 8 seconds was used between beginning of the contrast media infusion and start of the measurement. Fifty-five kidneys could be evaluated with a total of 62 arteries including seven accessory vessels. Four patients initially presented after nephrectomy, and one postoperative patient had undergone nephrectomy.

Data Postprocessing
For quantitative and semiquantitative image analysis, the MR angiograms were reconstructed on a satellite console in a standardized protocol. For initial evaluation, a targeted rectangular maximum intensity projection (MIP) image was obtained that encompassed the extra- and intrarenal portions of each renal artery. The MIP images were rotated in the axial and coronal plane with an increment of 15° between each image. In cases with severe kinking of the renal arteries or overlay of other vessel structures and for the analysis of intrarenal vessels, the target volume MIP images were complemented by means of multiplanar reconstruction of individual sections. These sections were then loaded into the MIP image program and used for subvolume double-oblique MIP images.

Image Analysis
Quantitative analysis.—For quantitative assessment of vascular enhancement, signal intensity measurements were performed on the standard 3D MR angiograms and on the individual phases obtained with the multiphase technique, including the nonenhanced, early arterial, main arterial, early venous, and, if available, late venous phase. The measurements were performed on the individual sections of the 3D volume set as well as on the multiplanar, reformatted, subvolume MIP images. To standardize the volume of the MIP images, a similar number of multiplanar reformatted base images were used as the data set in all patients. This data set always encompassed the aorta, one renal artery, and the central portions of one kidney. To control for variations in anatomic dimensions, the angle of the oblique 3D data volume was adjusted accordingly. The left and right renal arteries were analyzed separately. Regions of interest were placed in the aorta at the level of the renal arteries, in the proximal and distal renal artery, and intrarenally in a segmental vessel. The diameter of the regions of interest was adapted to the size of the vessel lumen and was always positioned in a nonstenotic section of each vessel segment. The middle segmental renal artery was chosen as the position of the intrarenal region of interest, since this vessel was identified most frequently. In case of poor depiction, a different segmental vessel that revealed the highest signal intensity was chosen. Vessel contrast-to-background ratios were calculated for all four vascular locations on the individual sections and on the oblique subvolume MIP images by dividing the signal intensity in the intravascular region of interest by the mean signal intensity in two paravascular regions of interest placed within 5 mm of the vessel margin.

Semiquantitative analysis.—For further assessment of image quality and vessel conspicuity, the multiplanar reconstructed images of the standard and multiphase 3D MR angiograms were independently analyzed by two radiologists (S.O.S., M.V.K.) who are experienced in vascular imaging. All images (ie, standard and multiphase MR angiograms) were jointly randomized and shown to the readers in order of randomization.

General criteria included the impression of overall image quality with use of a score of 1–5: 1, very good; 2, good; 3, average; 4, fair; or 5, unsatisfactory. Image degradation as a result of overlay of renal vein, overlay of accumulated contrast material from the test bolus in the renal pelvis, or breathing artifacts was rated as 1, none; 2, minor, not affecting a diagnostic evaluation; or 3, major, substantially affecting a diagnostic evaluation.

In addition, vessel conspicuity for the extra- and intrarenal vascular tree was evaluated at the level of the ostium, proximal two-thirds, distal one-third, and segmental vessels on the oblique subvolume MIP images. To avoid influences of the vessel diameter on the evaluation, only the main renal arteries (83 on standard images, 55 on multiphase images) were chosen for analysis of the extrarenal vascular tree. A score of 1–5 was used: 1, vessel segment not identified; 2, vaguely identified; 3, identified but poorly and incontinuously defined; 4, clearly and continuously identified but indefinite evaluation of patency; or 5, clearly identified with definite evaluation of patency.

The diagnostic confidence for grading a stenosis as either not present (0%), mild (<50%), moderate (50%–80%), severe (>80%), or occluded was rated as 1, definite; 2, highly probable; 3, probable; 4, uncertain; or 5, indefinite.

Correlation of Multiphase Angiography with Standard of Reference
For the multiphase angiograms, correlation with digital subtraction angiograms was possible in 17 patients. At DSA, a transfemorally inserted 5-F pigtail catheter was positioned at the level of the suspected renal artery origins between T12 and L1. The catheter position was verified by injecting 4 mL of nonionic contrast media (Imagopaque 300; Nycomed, Wayne, Pa). Anteroposterior DSA was performed with a C-arm system (Polydiagnost DVI-II; Philips Medical Systems, Shelton, Conn) by injecting 30 mL of the contrast media at a rate of 15 mL/sec. In 14 of the 17 patients, additional oblique views were obtained at an angle of 18°–25°. In case of a distal stenosis, the renal artery was also selectively catheterized with a 5-F double curve or cobra catheter, and an additional 9–10 mL of the contrast media was injected at a rate of 5–6 mL/sec. In an additional four patients, correlation with surgical findings was available. Intraoperatively, the stenosis was resected, and the renal artery was either reinserted into the aorta, or an end-to-end anastomosis was performed with or without interposition of a vein graft. For accurate determination of the degree of stenosis, the resected segment was filled with epoxy resin to obtain a cast of the lumen.

In these 21 patients, a total of 39 main renal arteries and four accessory renal arteries were present. Stenosis grading was performed independently by both readers, who were blinded to the DSA or surgical findings. Stenoses were graded with the grading scheme described previously. At DSA, two mild, six moderate, and seven severe stenoses of the main renal arteries were found. In the additional four patients with surgical correlation, two moderate and three severe main renal artery stenoses were identified intraoperatively. No accessory renal artery stenosis was present.

Statistical Analysis
Data entry procedures and statistical analysis were performed with a statistical software system (SAS for Windows, version 6.12; SAS Institute, Cary, NC). For statistical test comparisons, nonparametric methods were applied. Interval data arising from signal intensity measurements were analyzed with the Wilcoxon-Mann-Whitney U test, and ordinal data were analyzed with the Mantel-Haenszel test. For comparison of observer performance, without consensus, a statistic was used. The P values are exploratory in nature, and therefore no Bonferroni correction was made. Interobserver agreement was considered as slight, 0.2; fair, = 0.21–0.40; moderate, = 0.41–0.60; substantial, = 0.61–0.80; or almost perfect, = 0.81–1.00 (17).

	RESULTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
In all cases, imaging of the renal vasculature was successful with multiphase 3D MR angiography during arrival of the contrast media bolus, and no technical failures occurred. All images were of adequate quality. Mean image reconstruction time was about 3 minutes for the five 3D data sets. In all patients for each of the five phases, the short acquisition time of 6 seconds allowed reliable differentiation of the early and main arterial phases and of at least the early venous phase (Fig 2). With the imaging delay of 8 seconds between the beginning of contrast media injection and the start of multiphase imaging, one nonenhanced phase image was acquired in 21 of 30 (70%) studies, followed by images of the four different enhancement phases. In seven (23%) studies, bolus arrival was delayed, and two (five studies) or three (two studies) nonenhanced phase images were obtained prior to bolus arrival. In two (7%) studies, no nonenhanced phase image was acquired, but early arterial phase images were obtained in all studies.

View larger version (139K): [in this window] [in a new window]   Figure 2a. (a–c) MIP images obtained from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions) obtained in a 57-year-old man with bilateral proximal renal artery stenosis (right, moderate [50%–80%]; left, severe [>80%]). Images for five phases were obtained in a single breath hold with an acquisition time of 5.8 seconds per phase. Measurement was started 8 seconds after the beginning of the contrast media injection with an infusion rate of 3 mL/sec. (a) Early arterial phase MIP image shows intense enhancement of the proximal abdominal aorta and the complete renovascular tree with only minor renoparenchymal enhancement and no venous overlay. Note the clear depiction of the anterior and posterior division and middle segmental artery (arrow). Owing to the short acquisition time, no ringing artifacts are seen, even though the individual section was acquired during rapidly increasing concentrations of gadopentetate dimeglumine. (b) Main arterial phase MIP image (coronal view) reveals major parenchymal enhancement. Maximum vessel-to-background contrast is present in the aorta and extrarenal portions of the renal artery, but the intrarenal vessels (open arrow) are no longer seen. Minor venous overlay is caused by the distal renal veins (solid arrow), which further increases renal enhancement but does not substantially affect evaluation of the proximal extrarenal vascular tree. (c) Early venous phase MIP image (coronal view) shows enhancement of the complete renal vein up to the junction with the inferior vena cava (arrow), which substantially hinders analysis of the renal artery segments. Arterial enhancement is present, and other rotated MIP images (not shown) with better separation of arteries and veins could be potentially diagnostic for the complete extrarenal artery. No breathing artifacts are present on this fourth-phase image because it was acquired not more than 20 seconds after imaging was initiated. (d) DSA image (anteroposterior view).

View larger version (133K): [in this window] [in a new window]   Figure 2b. (a–c) MIP images obtained from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions) obtained in a 57-year-old man with bilateral proximal renal artery stenosis (right, moderate [50%–80%]; left, severe [>80%]). Images for five phases were obtained in a single breath hold with an acquisition time of 5.8 seconds per phase. Measurement was started 8 seconds after the beginning of the contrast media injection with an infusion rate of 3 mL/sec. (a) Early arterial phase MIP image shows intense enhancement of the proximal abdominal aorta and the complete renovascular tree with only minor renoparenchymal enhancement and no venous overlay. Note the clear depiction of the anterior and posterior division and middle segmental artery (arrow). Owing to the short acquisition time, no ringing artifacts are seen, even though the individual section was acquired during rapidly increasing concentrations of gadopentetate dimeglumine. (b) Main arterial phase MIP image (coronal view) reveals major parenchymal enhancement. Maximum vessel-to-background contrast is present in the aorta and extrarenal portions of the renal artery, but the intrarenal vessels (open arrow) are no longer seen. Minor venous overlay is caused by the distal renal veins (solid arrow), which further increases renal enhancement but does not substantially affect evaluation of the proximal extrarenal vascular tree. (c) Early venous phase MIP image (coronal view) shows enhancement of the complete renal vein up to the junction with the inferior vena cava (arrow), which substantially hinders analysis of the renal artery segments. Arterial enhancement is present, and other rotated MIP images (not shown) with better separation of arteries and veins could be potentially diagnostic for the complete extrarenal artery. No breathing artifacts are present on this fourth-phase image because it was acquired not more than 20 seconds after imaging was initiated. (d) DSA image (anteroposterior view).

View larger version (143K): [in this window] [in a new window]   Figure 2c. (a–c) MIP images obtained from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions) obtained in a 57-year-old man with bilateral proximal renal artery stenosis (right, moderate [50%–80%]; left, severe [>80%]). Images for five phases were obtained in a single breath hold with an acquisition time of 5.8 seconds per phase. Measurement was started 8 seconds after the beginning of the contrast media injection with an infusion rate of 3 mL/sec. (a) Early arterial phase MIP image shows intense enhancement of the proximal abdominal aorta and the complete renovascular tree with only minor renoparenchymal enhancement and no venous overlay. Note the clear depiction of the anterior and posterior division and middle segmental artery (arrow). Owing to the short acquisition time, no ringing artifacts are seen, even though the individual section was acquired during rapidly increasing concentrations of gadopentetate dimeglumine. (b) Main arterial phase MIP image (coronal view) reveals major parenchymal enhancement. Maximum vessel-to-background contrast is present in the aorta and extrarenal portions of the renal artery, but the intrarenal vessels (open arrow) are no longer seen. Minor venous overlay is caused by the distal renal veins (solid arrow), which further increases renal enhancement but does not substantially affect evaluation of the proximal extrarenal vascular tree. (c) Early venous phase MIP image (coronal view) shows enhancement of the complete renal vein up to the junction with the inferior vena cava (arrow), which substantially hinders analysis of the renal artery segments. Arterial enhancement is present, and other rotated MIP images (not shown) with better separation of arteries and veins could be potentially diagnostic for the complete extrarenal artery. No breathing artifacts are present on this fourth-phase image because it was acquired not more than 20 seconds after imaging was initiated. (d) DSA image (anteroposterior view).

View larger version (161K): [in this window] [in a new window]   Figure 2d. (a–c) MIP images obtained from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions) obtained in a 57-year-old man with bilateral proximal renal artery stenosis (right, moderate [50%–80%]; left, severe [>80%]). Images for five phases were obtained in a single breath hold with an acquisition time of 5.8 seconds per phase. Measurement was started 8 seconds after the beginning of the contrast media injection with an infusion rate of 3 mL/sec. (a) Early arterial phase MIP image shows intense enhancement of the proximal abdominal aorta and the complete renovascular tree with only minor renoparenchymal enhancement and no venous overlay. Note the clear depiction of the anterior and posterior division and middle segmental artery (arrow). Owing to the short acquisition time, no ringing artifacts are seen, even though the individual section was acquired during rapidly increasing concentrations of gadopentetate dimeglumine. (b) Main arterial phase MIP image (coronal view) reveals major parenchymal enhancement. Maximum vessel-to-background contrast is present in the aorta and extrarenal portions of the renal artery, but the intrarenal vessels (open arrow) are no longer seen. Minor venous overlay is caused by the distal renal veins (solid arrow), which further increases renal enhancement but does not substantially affect evaluation of the proximal extrarenal vascular tree. (c) Early venous phase MIP image (coronal view) shows enhancement of the complete renal vein up to the junction with the inferior vena cava (arrow), which substantially hinders analysis of the renal artery segments. Arterial enhancement is present, and other rotated MIP images (not shown) with better separation of arteries and veins could be potentially diagnostic for the complete extrarenal artery. No breathing artifacts are present on this fourth-phase image because it was acquired not more than 20 seconds after imaging was initiated. (d) DSA image (anteroposterior view).

View larger version (26K): [in this window] [in a new window]   Figure 3. Graphs depict comparison of vessel contrast-to-background ratio in the aorta and proximal, distal, and segmental renal artery on multiphase (solid line) and standard (dashed line) 3D MR angiograms obtained in the different phases () of contrast media passage. Findings are compared for A, individual sections (individual slices) and B, MIP (subvolume MIPs) images. SDs are indicated by error bars for the multiphase 3D MR angiograms and by dotted lines for the standard 3D MR angiograms. In the main arterial phase, increase in vessel contrast-to-background ratio in the distal and intrarenal vessels is substantially reduced as a result of overlay of renal parenchyma. A, Vessel contrast-to-background ratios on the multiphase 3D MR angiograms are less than those on the standard 3D MR angiograms for all vascular regions owing to higher background signal intensity. B, These differences are not present on the standard and multiphase subvolume MIP images. In the early arterial phase, contrast-to-background ratio of the segmental arteries on multiphase MIP images substantially exceeds that on standard MIP images, since no overlay of renal parenchyma has occurred.

Comparison of the early arterial multiphase and standard 3D MR angiograms by means of the Wilcoxon-Mann-Whitney U test revealed a P value of less than .001 for the subvolume MIP images of the intrarenal vessel segment (Table 1). No statistically significant differences were found between images of the extrarenal segments in the early arterial phase or for any of the vessels segments in the main arterial phase.

View larger version (117K): [in this window] [in a new window]   Figure 4a. Factors that influence distal vessel conspicuity on delayed standard 3D MR angiograms (5/2, 26 x 35-cm field of view, 9-cm coronal slab, 50 partitions) obtained with a 1-mL test bolus and a 32-second acquisition time. (a, b) Oblique subvolume MIP images encompass the aorta, right renal artery, and central portions of the kidney. (a) In a 62-year-old man, the proximal renal artery reveals adequate contrast enhancement. However, overlay of the enhanced renal vein (solid arrow) masks the anterior and posterior divisions. Owing to massive parenchymal enhancement, intrarenal vessels (open arrow) are not visible. (b) In a 54-year-old man, the image was perfectly timed. No venous overlay is present for the proximal and distal renal artery. Nonetheless, intrarenal vessels are not visible owing to parenchymal enhancement. In addition, accumulation of contrast media from the test bolus in the renal pelvis (solid arrow) further hampers the delineation of vessel structures.

View larger version (137K): [in this window] [in a new window]   Figure 4b. Factors that influence distal vessel conspicuity on delayed standard 3D MR angiograms (5/2, 26 x 35-cm field of view, 9-cm coronal slab, 50 partitions) obtained with a 1-mL test bolus and a 32-second acquisition time. (a, b) Oblique subvolume MIP images encompass the aorta, right renal artery, and central portions of the kidney. (a) In a 62-year-old man, the proximal renal artery reveals adequate contrast enhancement. However, overlay of the enhanced renal vein (solid arrow) masks the anterior and posterior divisions. Owing to massive parenchymal enhancement, intrarenal vessels (open arrow) are not visible. (b) In a 54-year-old man, the image was perfectly timed. No venous overlay is present for the proximal and distal renal artery. Nonetheless, intrarenal vessels are not visible owing to parenchymal enhancement. In addition, accumulation of contrast media from the test bolus in the renal pelvis (solid arrow) further hampers the delineation of vessel structures.

View larger version (120K): [in this window] [in a new window]   Figure 5a. MIP images from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.4-second acquisition time for each phase) obtained in a 58-year-old man suspected of having renal artery stenosis. In a–c, the solid arrow indicates a moderate stenosis of the left renal artery at the level of the early extrarenal bifurcation. (a) Early arterial phase, coronal MIP image shows no parenchymal enhancement. A right-side stenosis of the posterior segmental renal artery (open arrow) could not be ruled out. (b) Early arterial phase, axial-oblique MIP image clearly demonstrates patency (open arrow) of the right-side vascular segment in the nonenhanced surrounding parenchyma. (c) Main arterial phase, coronal MIP image shows delineation of the distal (open arrow) and intrarenal vessel segments is substantially limited by major parenchymal enhancement. (d) Main arterial phase, axial-oblique MIP image shows that evaluation of definite right-side patency (open arrow) is difficult.

View larger version (109K): [in this window] [in a new window]   Figure 5c. MIP images from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.4-second acquisition time for each phase) obtained in a 58-year-old man suspected of having renal artery stenosis. In a–c, the solid arrow indicates a moderate stenosis of the left renal artery at the level of the early extrarenal bifurcation. (a) Early arterial phase, coronal MIP image shows no parenchymal enhancement. A right-side stenosis of the posterior segmental renal artery (open arrow) could not be ruled out. (b) Early arterial phase, axial-oblique MIP image clearly demonstrates patency (open arrow) of the right-side vascular segment in the nonenhanced surrounding parenchyma. (c) Main arterial phase, coronal MIP image shows delineation of the distal (open arrow) and intrarenal vessel segments is substantially limited by major parenchymal enhancement. (d) Main arterial phase, axial-oblique MIP image shows that evaluation of definite right-side patency (open arrow) is difficult.

View larger version (111K): [in this window] [in a new window]   Figure 5b. MIP images from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.4-second acquisition time for each phase) obtained in a 58-year-old man suspected of having renal artery stenosis. In a–c, the solid arrow indicates a moderate stenosis of the left renal artery at the level of the early extrarenal bifurcation. (a) Early arterial phase, coronal MIP image shows no parenchymal enhancement. A right-side stenosis of the posterior segmental renal artery (open arrow) could not be ruled out. (b) Early arterial phase, axial-oblique MIP image clearly demonstrates patency (open arrow) of the right-side vascular segment in the nonenhanced surrounding parenchyma. (c) Main arterial phase, coronal MIP image shows delineation of the distal (open arrow) and intrarenal vessel segments is substantially limited by major parenchymal enhancement. (d) Main arterial phase, axial-oblique MIP image shows that evaluation of definite right-side patency (open arrow) is difficult.

View larger version (94K): [in this window] [in a new window]   Figure 5d. MIP images from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.4-second acquisition time for each phase) obtained in a 58-year-old man suspected of having renal artery stenosis. In a–c, the solid arrow indicates a moderate stenosis of the left renal artery at the level of the early extrarenal bifurcation. (a) Early arterial phase, coronal MIP image shows no parenchymal enhancement. A right-side stenosis of the posterior segmental renal artery (open arrow) could not be ruled out. (b) Early arterial phase, axial-oblique MIP image clearly demonstrates patency (open arrow) of the right-side vascular segment in the nonenhanced surrounding parenchyma. (c) Main arterial phase, coronal MIP image shows delineation of the distal (open arrow) and intrarenal vessel segments is substantially limited by major parenchymal enhancement. (d) Main arterial phase, axial-oblique MIP image shows that evaluation of definite right-side patency (open arrow) is difficult.

View larger version (140K): [in this window] [in a new window]   Figure 6a. Images obtained in a 56-year-old man with acute dissection of the left renal artery, which resulted in a long renal artery stenosis and a poststenotic aneurysm. (a) Left oblique selective DSA image reveals a large aneurysm (open arrow) that involves the distal renal artery with a proximal stenotic segment (solid arrow). (b–d) MIP images from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.4-second acquisition time for each phase) confirm the DSA findings. In b and c, early arterial phase images show only minor renal parenchymal enhancement, which allows identification of the distal end of the aneurysm (solid arrow in c) prior to branching into the lobar arteries. (b) Coronal MIP image shows the stenotic segment of the main renal artery (solid arrow) and an aneurysm of the distal portion (open arrow). (c) Coronal-oblique MIP image shows additional involvement of the posterior division or posterior segmental artery by the aneurysm (open arrow) that was masked in a by overlay of multiple vessels but was confirmed intraoperatively. (d) On late arterial phase MIP image, rotated as in c, further aneurysm extension (open arrow) cannot be definitely ruled out. The extrarenal portion of an accessory renal artery is clearly depicted, but the distal portion is depicted less clearly than in b and c. A renal infarct (solid arrow) in the lateral portion resulted from recurrent thrombotic events. Delayed and decreased enhancement is present in the different phases.

View larger version (119K): [in this window] [in a new window]   Figure 6b. Images obtained in a 56-year-old man with acute dissection of the left renal artery, which resulted in a long renal artery stenosis and a poststenotic aneurysm. (a) Left oblique selective DSA image reveals a large aneurysm (open arrow) that involves the distal renal artery with a proximal stenotic segment (solid arrow). (b–d) MIP images from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.4-second acquisition time for each phase) confirm the DSA findings. In b and c, early arterial phase images show only minor renal parenchymal enhancement, which allows identification of the distal end of the aneurysm (solid arrow in c) prior to branching into the lobar arteries. (b) Coronal MIP image shows the stenotic segment of the main renal artery (solid arrow) and an aneurysm of the distal portion (open arrow). (c) Coronal-oblique MIP image shows additional involvement of the posterior division or posterior segmental artery by the aneurysm (open arrow) that was masked in a by overlay of multiple vessels but was confirmed intraoperatively. (d) On late arterial phase MIP image, rotated as in c, further aneurysm extension (open arrow) cannot be definitely ruled out. The extrarenal portion of an accessory renal artery is clearly depicted, but the distal portion is depicted less clearly than in b and c. A renal infarct (solid arrow) in the lateral portion resulted from recurrent thrombotic events. Delayed and decreased enhancement is present in the different phases.

View larger version (137K): [in this window] [in a new window]   Figure 6c. Images obtained in a 56-year-old man with acute dissection of the left renal artery, which resulted in a long renal artery stenosis and a poststenotic aneurysm. (a) Left oblique selective DSA image reveals a large aneurysm (open arrow) that involves the distal renal artery with a proximal stenotic segment (solid arrow). (b–d) MIP images from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.4-second acquisition time for each phase) confirm the DSA findings. In b and c, early arterial phase images show only minor renal parenchymal enhancement, which allows identification of the distal end of the aneurysm (solid arrow in c) prior to branching into the lobar arteries. (b) Coronal MIP image shows the stenotic segment of the main renal artery (solid arrow) and an aneurysm of the distal portion (open arrow). (c) Coronal-oblique MIP image shows additional involvement of the posterior division or posterior segmental artery by the aneurysm (open arrow) that was masked in a by overlay of multiple vessels but was confirmed intraoperatively. (d) On late arterial phase MIP image, rotated as in c, further aneurysm extension (open arrow) cannot be definitely ruled out. The extrarenal portion of an accessory renal artery is clearly depicted, but the distal portion is depicted less clearly than in b and c. A renal infarct (solid arrow) in the lateral portion resulted from recurrent thrombotic events. Delayed and decreased enhancement is present in the different phases.

View larger version (121K): [in this window] [in a new window]   Figure 6d. Images obtained in a 56-year-old man with acute dissection of the left renal artery, which resulted in a long renal artery stenosis and a poststenotic aneurysm. (a) Left oblique selective DSA image reveals a large aneurysm (open arrow) that involves the distal renal artery with a proximal stenotic segment (solid arrow). (b–d) MIP images from multiphase 3D MR angiograms (3.2/1.1, 26 x 35-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.4-second acquisition time for each phase) confirm the DSA findings. In b and c, early arterial phase images show only minor renal parenchymal enhancement, which allows identification of the distal end of the aneurysm (solid arrow in c) prior to branching into the lobar arteries. (b) Coronal MIP image shows the stenotic segment of the main renal artery (solid arrow) and an aneurysm of the distal portion (open arrow). (c) Coronal-oblique MIP image shows additional involvement of the posterior division or posterior segmental artery by the aneurysm (open arrow) that was masked in a by overlay of multiple vessels but was confirmed intraoperatively. (d) On late arterial phase MIP image, rotated as in c, further aneurysm extension (open arrow) cannot be definitely ruled out. The extrarenal portion of an accessory renal artery is clearly depicted, but the distal portion is depicted less clearly than in b and c. A renal infarct (solid arrow) in the lateral portion resulted from recurrent thrombotic events. Delayed and decreased enhancement is present in the different phases.

View larger version (117K): [in this window] [in a new window]   Figure 7a. MIP images from multiphase 3D MR angiograms (3.2/1.1, 30 x 34-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.8-second acquisition time for each phase) were obtained in a 53-year-old woman with fibromuscular dysplasia and a distal renal artery stenosis. (a) Early arterial phase, coronal MIP image clearly depicts a high-grade stenosis in the distal portion of the right renal artery (arrow). The stenosis is seen on the medial portion of the renal parenchyma with the presence of only minor background enhancement. The stenosis was confirmed intraoperatively. (b) On the late arterial phase, coronal MIP image, detection of the stenosis is substantially affected by the high signal intensity of the surrounding renal parenchyma (arrow). In contrast, evaluation of the proximal renal artery segments is possible.

View larger version (121K): [in this window] [in a new window]   Figure 7b. MIP images from multiphase 3D MR angiograms (3.2/1.1, 30 x 34-cm field of view, 8-cm coronal slab, 44 reconstructed partitions, 6.8-second acquisition time for each phase) were obtained in a 53-year-old woman with fibromuscular dysplasia and a distal renal artery stenosis. (a) Early arterial phase, coronal MIP image clearly depicts a high-grade stenosis in the distal portion of the right renal artery (arrow). The stenosis is seen on the medial portion of the renal parenchyma with the presence of only minor background enhancement. The stenosis was confirmed intraoperatively. (b) On the late arterial phase, coronal MIP image, detection of the stenosis is substantially affected by the high signal intensity of the surrounding renal parenchyma (arrow). In contrast, evaluation of the proximal renal artery segments is possible.

	DISCUSSION

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Besides being limited by imaging time and spatial resolution, breath-hold 3D MR angiography of the renal arteries is also limited by adequate bolus timing with potential overlay of multiple successively enhancing structures, including renal parenchyma and intra- and extrarenal veins. In this study, two imaging sequences were compared that have comparable spatial resolution but substantially different acquisition times for one 3D data set. We found that fast acquisition of multiple 3D data sets allowed arterial phase imaging of the renal arteries without prior bolus timing.

Imaging during the first transit of extracellular contrast agents is essential for short repetition time contrast-enhanced 3D MR angiography, since sufficient T1 shortening for adequate vessel-to-background ratio occurs only during peak blood concentrations of standard gadolinium chelates. With use of common gadolinium chelate doses of 0.2 mmol/kg and injection rates of 2–3 mL/sec, intraarterial peak concentrations last only 10–20 seconds, depending on the amount of saline solution flush, cardiac output, and type of contrast agent. Standard 3D sequences require use of imaging times of about 30 seconds for appropriate spatial resolution and anatomic coverage.

Since contrast information is determined in the low spatial frequencies, acquisition of the central third of k space (ie, 10 seconds of the total imaging time) has to be synchronized with the short plateau of the gadolinium chelate enhancement peak. In the approach used frequently for bolus timing, a test-bolus image is obtained to calculate contrast material arrival time (9–11). We used this approach in our standard 3D MR angiographic sequence. Although this technique can be readily performed, accumulation of contrast media from the test bolus in the excretory system after only a few minutes substantially affects the delineation of intrarenal vessels on a standard 3D MR angiogram (Table 2, Fig 4b). This occurs with the smallest gadolinium chelate dose of 1 mL.

Our fast multiphase acquisition did not require a test bolus; therefore, these diagnostic limitations were not present (Fig 2a). Despite use of a test bolus in our standard 3D MR angiographic sequence, the overall quality of enhancement was significantly lower than that with the multiphase technique (Table 2). This might be related to the difference in enhancement kinetics of the test bolus compared to those with a full contrast media dose. Improper start of the sequence due to operator or system delays may also contribute. Our multiphase protocol resulted in satisfactory enhancement in all cases. The 8-second imaging delay after start of the contrast media injection proved to be a reliable and robust way to minimize imaging time prior to contrast media arrival without missing the bolus (Fig 2a–2c). Enhancement of intrarenal arteries was rapidly followed by parenchymal enhancement after only 2–3 seconds (Fig 1), which was seen as overlay on MIP images (Figs 2b, 4). Intrarenal veins enhance only 5 seconds later. This enhancement may occur during acquisition of the periphery of k space in a 30-second standard acquisition, causing edge enhancement of the veins on the images. Multiple fast (6-second) 3D acquisitions allow imaging of an early arterial phase of the renal vasculature with minimal parenchymal and no venous enhancement (Figs 2a, 5a). Consequently, both readers considered the effect of image degradation in the early arterial phase as a result of overlay of renal vein or renal pelvis lower than that at standard 3D MR angiography (Table 2). Interobserver agreement was substantial (Table 4).

On the subvolume MIP images from early arterial phase 3D MR angiograms, vessel-to-background contrast of the segmental arteries was higher than that on the standard 3D MR angiograms (Table 1). These differences were not present in the analysis of the individual sections because the extraparenchymal intrarenal arteries are inconsistently attached to other enhancing structures such as renal parenchyma or veins. Therefore, on each section, contrast-to-background ratios were highly dependent on the location of the region of interest. Subvolume MIP images were found to be more representative, since local signal intensity differences average out on the projection images. In addition, only the subvolume MIP images allowed continuous analysis of the vessel course sufficient for the diagnosis of vessel patency. On the early arterial phase images, both readers found markedly improved distal vessel conspicuity (Table 3). In particular, this was true for torturous divisions and segmental arteries that could be continuously identified on multiple rotations of the subvolume MIP images owing to reduced parenchymal enhancement (Figs 2, 5, 6).

For the segmental vessels, however, the mode scores for both readers reached a diagnostic grade (grade 5) for only the middle segmental artery. In addition, the interobserver variability was only fair in some cases (Table 4). This is primarily explained by the difficult assessment of vessels that extend into the inhomogeneous margins of the 3D slab. Finally, even in a 6-second acquisition window, moderate parenchymal enhancement was present in some cases with a short contrast media transit time. The results might also be biased because two different populations were analyzed for comparison.

In this study, the spatial resolution for standard 3D MR angiography was comparable to that in the current literature for breath-hold imaging (10–13,18,19). At multiphase 3D MR angiography, this spatial resolution could almost be equaled by using the following features: (a) with the gradient overdrive, short repetition times of 3.2 msec could be achieved, and (b) with asymmetric k-space acquisition combined with zero filling, a substantial reduction in the acquisition time could be achieved. Asymmetric k-space acquisitions proved to be reliable imaging strategies with only minor ringing artifacts, and resolution is comparable to that on full k-space images. Therefore, both techniques could be compared with respect to differences in vessel conspicuity. In several studies, high sensitivities and specificities greater than 90% are reported for detection of proximal stenoses with standard test-bolus 3D MR angiography performed with parameters comparable to those used in this study (10,11,20,21). Since spatial resolution with the multiphase technique was comparable to that with the standard technique, with better image quality and less image degradation, comparable results are expected for grading of proximal stenoses. All proximal stenoses with DSA or surgical correlation were identified by reader 1, and only two were incorrectly graded by both readers (Table 5). Interobserver agreement was almost perfect.

On MR images, Prince et al (18) found the enhancement pattern of kidneys with hemodynamically significant stenoses was significantly different from that on the nonstenotic side. Multiple 3D data sets with high temporal resolution depict this phenomenon more clearly than do standard single-phase 3D MR angiograms obtained with longer acquisition times (25) (Fig 6). Changes in renal enhancement kinetics also occur with different perfusion rates in cases such as aortic dissection if the renal arteries arise from the true and false lumen. Since different phases of the contrast media transit are acquired in one image during a single breath hold, the anatomic information about vascular structures with different temporal enhancement kinetics can be combined in one image by adding multiple data sets together. Subtraction of the nonenhanced phase data further improves the contrast-to-noise ratio.

Acquisition of central k space during rapidly changing gadolinium chelate concentrations is known to cause typical stringlike ringing artifacts (12,26). During fast multiphasic acquisition, the relative temporal variation in the gadolinium chelate concentration is much smaller for each phase; therefore, no ringing artifacts were present on any of the multiphase 3D MR angiograms. In addition, none of the multiphase studies revealed breathing artifacts that affected image analysis substantially, since all patients were at least able to hold their breath for the first three phases. More breathing artifacts were found on standard 3D MR angiograms, but this difference was only mild because centric reordered acquisitions are less susceptible to an incomplete breath hold (16).

Other 3D MR angiographic strategies are being pursued that do not require a test bolus, but so far they are limited to certain types of MR systems (12,13). Our multiphase approach requires only fast gradients, which are available from various manufacturers. In addition, use of more efficient k-space sampling strategies, such as half Fourier (27), might allow implementation of multiphase 3D MR angiography on systems with lower gradient performance.

Multiphase 3D MR angiography is a convenient and reliable technique for arterial phase angiography of the renal arteries that is not susceptible to artifacts due to overlay of enhanced renal vein or pelvis. With comparable spatial resolution, depiction of hilar and larger intrarenal arteries is improved compared to that with standard 3D MR angiographic techniques requiring imaging times longer than 20 seconds. Fast acquisition without bolus timing procedures reduces total magnet time and facilitates a cost-effective combination with functional evaluations such as MR flow measurements in the renal artery (28,29).

	Acknowledgments

 
The authors thank Peter Miltner, MD, for performing the DSA examinations in the majority of patients, Stefan Groeninger, MS, and Mathias Blasche, MS, for technical support, Albrecht Moeller, PhD, (Medidata) and Nils Dannenberg, RT, for assistance with the data analysis, and Josef Wiegand and Christian Konold for photography.

	Footnotes

 
See also the editorial by Tello and Ptak (pp 605–607 ) in this issue.

Abbreviations: DSA = digital subtraction angiography FLASH = fast low-angle shot MIP = maximum intensity projection 3D = three-dimensional

Author contributions: Guarantors of integrity of entire study, S.O.S., M.V.K.; study concepts and design, S.O.S., M.V.K.; definition of intellectual content, S.O.S., M.B.; literature research, S.O.S.; clinical studies, S.O.S., M.E., F.K., H.H., M.V.K.; experimental studies, M.B., G.L.; data acquisition, S.O.S., M.B.; data analysis, S.O.S., M.V.K., M.E.; statis