Simultaneous static and cine nonenhanced MR angiography using radial sampling and highly constrained back projection reconstruction.

PURPOSE: To describe a pulse sequence for simultaneous static and cine nonenhanced magnetic resonance angiography (NEMRA) of the peripheral arteries. METHODS: The peripheral arteries of 10 volunteers and 6 patients with peripheral arterial disease (PAD) were imaged with the proposed cine NEMRA sequence on a 1.5 Tesla (T) system. The impact of multi-shot imaging and highly constrained back projection (HYPR) reconstruction was examined. The propagation rate of signal along the length of the arterial tree in the cine nonenhanced MR angiograms was quantified. RESULTS: The cine NEMRA sequence simultaneously provided a static MR angiogram showing vascular anatomy as well as a cine display of arterial pulse wave propagation along the entire length of the peripheral arteries. Multi-shot cine NEMRA improved temporal resolution and reduced image artifacts. HYPR reconstruction improved image quality when temporal reconstruction footprints shorter than 100 ms were used (P < 0.001). Pulse wave propagation within the arterial tree as displayed by cine NEMRA was slower in patients with PAD than in volunteers. CONCLUSION: Simultaneous static and cine NEMRA of the peripheral arteries is feasible. Multi-shot acquisition and HYPR reconstruction can be used to improve arterial conspicuity and temporal resolution.

Comparison of quiescent inflow single-shot and native space for nonenhanced peripheral MR angiography.

PURPOSE: To evaluate two nonenhanced MRA methods: quiescent-interval single-shot (QISS) and Native SPACE (NATIVE = Non-contrast Angiography of the Arteries and Veins; SPACE = Sampling Perfection with Application Optimized Contrast by using different flip angle Evolution), using contrast-enhanced MR angiography (CEMRA) as a reference standard. MATERIALS AND METHODS: Twenty patients (14 male; mean, 69.3 years old) referred for lower extremity MRA were recruited in a HIPAA-compliant prospective study. QISS and Native SPACE of the lower extremities were performed at 1.5 Tesla with a hybrid dual-injection contrast-enhanced MRA as reference. Image quality and stenosis severity were assessed in segments by two blinded radiologists. Methods were compared with logistic regression for correlated data for diagnostic accuracy. RESULTS: Of 496 arterial segments, 24 were considered nondiagnostic on the Native SPACE images. There were no QISS or CEMRA imaging segments considered to be nondiagnostic. Image quality was significantly higher for QISS than for Native SPACE. QISS stenosis sensitivity (84.9%) was not significantly different from Native SPACE (87.3%). QISS had better specificity (95.6%) than Native SPACE (87.0%), P = 0.0041. In comparison with QISS, Native SPACE proved less robust for imaging of the abdominal and pelvic segments. CONCLUSION: Native SPACE and QISS were sensitive for hemodynamically significant stenosis in this pilot study. QISS NEMRA demonstrated superior specificity and image quality, and was more robust in the abdominal and pelvic regions.

Evaluation of peripheral arterial disease with nonenhanced quiescent-interval single-shot MR angiography.

PURPOSE: To assess the diagnostic performance of quiescent-interval single-shot (QISS) magnetic resonance (MR) angiography, a nonenhanced two-dimensional electrocardiographically gated single-shot balanced steady-state free precession examination for the evaluation of symptomatic chronic lower limb ischemia. MATERIALS AND METHODS: For this prospective institutional review board-approved, HIPAA-compliant study, the institutional review board waived the requirement for informed patient consent. The QISS nonenhanced MR angiography technique was evaluated in a two-center trial involving 53 patients referred for lower extremity MR angiography for suspected or known chronic peripheral arterial disease (PAD), with contrast material-enhanced MR angiography serving as the noninvasive reference standard. The accuracy of stenosis assessments performed with the nonenhanced MR angiography sequence was evaluated relative to the reference standard. Per-segment, per-region, and per-limb sensitivities and specificities were calculated, and assessments were considered correct only if they were in exact agreement with the reference standard-derived assessments. Generalized estimating equation (GEE) modeling with use of an unstructured binomial logit analysis was used to account for clustering of multiple measurements per case. The sensitivity and specificity of QISS MR angiography for the determination of nonsignificant (<50%) versus significant (50%-100%) stenosis were compared with the sensitivity and specificity of the reference standard. RESULTS: The diagnostic performance of nonenhanced MR angiography was found to be nearly equivalent to the diagnostic performances of contrast-enhanced MR angiography and digital subtraction angiography. Non-GEE segment-based analysis revealed that for the two reviewers, nonenhanced MR angiography had sensitivities of 89.7% (436 of 486 segments) and 87.0% (423 of 486 segments) and specificities of 96.5% (994 of 1030 segments) and 94.6% (973 of 1028 segments). CONCLUSION: QISS nonenhanced MR angiography offers an alternative to currently used imaging tests for symptomatic chronic lower limb ischemia, for which the administration of iodinated or gadolinium-based contrast agents is contraindicated.

Highly accelerated contrast-enhanced MR angiography: improved reconstruction accuracy and reduced noise amplification with complex subtraction.

Contrast-enhanced magnetic resonance angiography is routinely performed using parallel imaging to best capture the first pass of contrast material through the target vasculature, followed by digital subtraction to suppress the appearance of unwanted signal from background tissue. Both processes, however, amplify noise and can produce uninterpretable images when large acceleration factors are used. Using a phantom study of contrast-enhanced magnetic resonance angiography, we show that complex subtraction processing prior to partially parallel reconstruction improves reconstruction accuracy relative to magnitude subtraction processing for reduction factors as large as 12. Time-resolved contrast-enhanced magnetic resonance angiographic data obtained with complex subtraction in volunteers supported the results of the phantom study and when compared with magnitude subtraction processing demonstrated reduced geometry factors as well as improved image quality at large reduction factors.

Quiescent-interval single-shot unenhanced magnetic resonance angiography of peripheral vascular disease: Technical considerations and clinical feasibility.

We performed technical optimization followed by a pilot clinical study of quiescent-interval single-shot MR angiography for peripheral vascular disease. Quiescent-interval single-shot MR angiography acquires data using a modified electrocardiographic (ECG)-triggered, fat suppressed, two-dimensional, balanced steady-state, free precession pulse sequence incorporating slice-selective saturation and a quiescent interval for maximal enhancement of inflowing blood. Following optimization at 1.5 T, a pilot study was performed in patients with peripheral vascular disease, using contrast-enhanced MR angiography as the reference standard. The optimized sequence used a quiescent interval of 228 ms, alpha/2 catalyzation of the steady-state magnetization, and center-to-out partial Fourier acquisition with parallel acceleration factor of 2. Spatial resolution was 2-3mm along the slice direction and 0.7-1mm in-plane before interpolation. Excluding stented arterial segments, the sensitivity, specificity, and positive and negative predictive values of quiescent-interval single-shot MR angiography for arterial narrowing greater than 50% or occlusion were 92.2%, 94.9%, 83.9%, and 97.7%, respectively. Quiescent-interval single-shot MR angiography provided robust depiction of normal peripheral arterial anatomy and peripheral vascular disease in less than 10 min, without the need to tailor the technique for individual patients. Moreover, the technique provides consistent image quality in the pelvic region despite the presence of respiratory and bowel motion.

Sub-millimeter isotropic MRI for segmentation of subcortical brain regions and brain visualization.

PURPOSE: To evaluate a rapid sub-millimeter isotropic spoiled gradient-echo (nonselective SPGR) to facilitate the brain subcortical segmentation and the visualization of brain volume compared with the commonly accepted inversion recovery-prepared SPGR (SPGR-IR) technique. MATERIALS AND METHODS: The feasibility of the nonselective SPGR was evaluated for two segmentation algorithms. FAST was used to segment the brain into constituent tissue classes (white matter, gray matter, cerebrospinal fluid) and FreeSurfer was used to segment specific subcortical structures (hippocampus, caudate, putamen, and thalamus). Localized apparent signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) values for nonselective SPGR and the SPGR-IR were compared for the studied subcortical regions. The three-dimensional volume rendering was generated to evaluate the nonselective SPGR and the SPGR-IR for brain visualization. RESULTS: In basal ganglia regions, nonselective SPGR allows for consistent segmentation results for both FAST and FreeSurfer. This sequence also better differentiated gray/white matter compared with SPGR-IR. An approximate two-fold improvement of image quality in apparent SNR and CNR was indicated for subcortical brain anatomical structures with nonselective SPGR versus SPGR-IR. The nonselective SPGR improved clarity and yielded a more realistic depiction of the brain surface for visualization compared with SPGR-IR. CONCLUSION: Compared with SPGR-IR, nonselective SPGR allows for consistent segmentation results for basal ganglia regions and improved clarity for visualization of the brain.

Cerebral venography using fluid-suppressed STARFIRE.

Fluid-suppressed STARFIRE (Signal Targeting Alternative Radiofrequency and Flow-Independent Relaxation Enhancement) is a noncontrast method for flow-independent MR venography (MRV). It uses magnitude subtraction of two inversion recovery-prepared segmented three-dimensional (3D) balanced steady-state free precession acquisitions to obtain isotropic cerebral venograms in which both fat and cerebrospinal fluid signals are suppressed. Unlike two-dimensional time-of-flight (2D TOF) MRV, it is insensitive to the flow velocity of the cerebral veins. The method provided excellent depiction of the dural venous sinuses and cortical veins on maximum intensity projection images. Fluid-suppressed STARFIRE and 2D TOF were compared with contrast-enhanced 3D MRV as the reference standard in seven healthy subjects at 1.5 Tesla. Fluid-suppressed STARFIRE compared favorably to 2D TOF on both quantitative and qualitative analyses. Contrast-enhanced MRV provided the highest vein-background relative contrast and best demonstrated the straight sinus, whereas STARFIRE depicted the most venous branches. Further investigation will be required to determine the accuracy for cerebral venous thrombosis.

Effect of nitric oxide synthase inhibition on intrarenal oxygenation as evaluated by blood oxygenation level-dependent magnetic resonance imaging.

OBJECTIVE: To investigate the feasibility of studying renal effects of nitric oxide synthase inhibition (NOSi) in humans by blood oxygenation level-dependent (BOLD) MRI. Nitric oxide (NO) is known to play a key role in the pathophysiology of hypertension and previous reports suggest reduced bioavailability of NO in the kidneys of hypertensive rats and hence show reduced response to NOSi using BOLD MRI. Ability to perform similar studies in humans could potentially lead to detection of early changes before development of symptoms, and to monitor novel interventions targeted toward improved NO bioavailability. The specific goals for this study were: (1) to examine whether lower doses and dose rate of administration of NOSi such as those previously used in humans can be detected by BOLD MRI in rat kidneys, (2) to compare changes in R2* to direct measures of renal medullary oxygen levels and blood flow using invasive probes (OxyLite/OxyFlo), and (3) to examine for the first time the effect of NOSi on intrarenal oxygenation in humans. MATERIAL AND METHODS: In rat kidneys, acute changes in renal tissue oxygenation induced by different doses (2, 4, and 10 mg/kg) of N-nitro-L-arginine methyl ester were studied in 36 Sprague Dawley rats, which were equally divided into BOLD MRI and OxyLite/OxyFlo groups. Similarly in humans, acute changes in renal oxygenation were induced by 2 different NOS inhibitors NG-monomethyl-L-arginine (4.25 mg/kg) in 7 volunteers and N-nitro-L-arginine methyl ester (2 mg/kg and 4 mg/kg) in 6 healthy young volunteers. A multiple gradient echo sequence was used in both rats (TE = 4.4-57.8 milliseconds with 3.6 milliseconds interecho spacing) and humans (TE = 6.4-40.8 milliseconds with a 2.3 milliseconds interecho spacing) to acquire 16 T2*-weighted images. R2* maps were constructed by fitting a single exponential decay to the image data on pixel by pixel basis. R2* measurements in the cortex and medulla were performed by regions of interest analysis. Measurements were performed before and during infusion of NOSi. RESULTS: In rats, NOSi decreased medullary pO2 and blood flow in a dose-dependent manner, and BOLD MRI showed an increase in medullary R2* consistent with the invasive pO2 measurements. In humans, BOLD MRI similarly showed an increase in medullary and cortical R2* after NOSi in a dose-dependent manner. In both rats and humans, the R2* values fell back toward baseline before the end of the infusion period. CONCLUSION: Comparison of BOLD MRI measurements with those using invasive probes suggests that changes in blood flow are at least partly responsible for observed changes with BOLD MRI. Monitoring changes after NOSi by renal BOLD MRI in vivo in human kidneys are feasible, and preliminary findings are consistent with observations in rat kidneys. Future studies are warranted to fully understand the apparent reversal in R2* changes during the infusion of NOSi.

Rapid whole-brain magnetic resonance imaging with isotropic resolution at 3 Tesla.

Isotropic imaging offers the potential of improving lesion detection and imaging efficiency by enabling orthogonal image reformations without loss of spatial resolution. However, lengthy scan times for T1-weighted isotropic data acquisitions have been an impediment to the routine clinical application of this approach. We tested the feasibility of using the improved signal-to-noise ratio at 3 Tesla to perform rapid, whole-brain T1-weighted imaging with isotropic 0.8 mm x 0.8 mm x 0.8 mm (0.51 mm3) voxels. The method was validated in healthy volunteers and patients.Eight healthy subjects were imaged pre- and postcontrast on a 3 Tesla MR system. T1-weighted, 3-dimensional spoiled gradient-echo (3D SPGR) data were acquired at 0.8-mm slice thickness and reconstructed at 2-mm thickness in 3 orthogonal orientations. Scan time was 4 minutes 42 seconds. The technique was compared with inversion recovery-prepared spoiled gradient-echo (SPGR-IR) and 2D spin-echo (SE) for comparable spatial resolution and scan time. It was then tested in comparison with 2D SE in a series of 10 patients with enhancing brain lesions.The 3D SPGR technique provided approximately twice the contrast-to-noise ratio (CNR) of SPGR-IR and 50% greater CNR than 2D SE for discriminating gray and white matter. Image quality ratings, evaluated with nonparametric analyses, were also significantly higher for 3D SPGR in both volunteers (P < 0.05) and patients with enhancing lesions (P < 0.05). Of particular note was the elimination of postcontrast vessel pulsation artifacts, which were commonly present with 2D SE, and more uniform hypointensity of cerebrospinal fluid. Lesion enhancement in patients did not differ significantly for 2D SE and 3D SPGR.Our results demonstrate the feasibility of rapid, whole-brain isotropic imaging at 3 Tesla using submillimeter voxels. Artifacts were minimal, especially compared with 2D SE, whereas CNR was 2-fold better than SPGR-IR. The capability for creating reformatted images in orthogonal orientations from a single isotropic acquisition greatly improves efficiency compared with 2D acquisitions acquired in multiple planes. Although further clinical study is needed in a larger patient cohort, these initial results suggest substantial clinical promise for the technique.

Gadolinium-enhanced off-resonance contrast angiography.

We describe a novel physical basis and methodology for gadolinium (Gd)-enhanced MRA, which we call "off-resonance contrast angiography" (ORCA). Unlike standard contrast-enhanced (CE) MR angiography (MRA), ORCA contrast depends not on T(1) but on Gd-induced shifts in intravascular resonance frequency due to the bulk magnetic susceptibility (BMS) effects of Gd. The method was tested at 3 Tesla in phantoms with a range of dilutions of Gd-DTPA and ultrasmall iron oxide contrast agent (CA). With the use of ORCA, complete background suppression was obtained without image subtraction. As a result, catheters filled with various Gd dilutions proved to be highly conspicuous in ORCA projection images. This feature may make ORCA particularly attractive for passive catheter tracking during MR-guided endovascular procedures. Gd-induced intravascular frequency shifts were measured in human subjects and found to be in the expected range. ORCA was used to create angiograms of forearm veins that were comparable in quality to standard CE-MRA. In addition, ORCA images of the extracranial carotid bifurcation were successfully acquired during intravenous contrast administration. However, significant technical restrictions also exist, including a dependence on vessel orientation with respect to B(0), and sensitivity to static field inhomogeneities. Further study is needed to determine the practicality and potential clinical utility of this method.

Lower extremity deep venous thrombosis: evaluation with ferumoxytol-enhanced MR imaging and dual-contrast mechanism--preliminary experience.

Institutional review board approval and informed consent were obtained for this HIPAA-compliant study, whose purpose was to prospectively evaluate the use of a dual-contrast mechanism in conjunction with an iron oxide blood pool contrast agent, ferumoxytol, to depict deep venous thrombosis (DVT). Nine patients with lower extremity DVT detected with duplex ultrasonography (US) were imaged with magnetic resonance (MR) imaging and ferumoxytol. Three techniques, including precontrast two-dimensional time-of-flight (TOF) imaging, ferumoxytol-enhanced bright-blood imaging, and ferumoxytol-enhanced dark-blood imaging, were applied. Image quality for precontrast and ferumoxytol-enhanced images was analyzed by using a four-point scale. Thrombus was depicted as a filling defect within the blood pool on bright-blood images and as bright tissue that appeared highly contrasted against a dark background on dark-blood images. Image quality of ferumoxytol-enhanced images was uniformly superior to that of precontrast TOF images (P = .007). Compared with precontrast TOF images, ferumoxytol-enhanced bright-blood images had higher contrast-to-noise ratios (CNRs) between thrombus and blood (P = .051), whereas ferumoxytol-enhanced dark-blood images showed significantly higher CNRs between thrombus and surrounding muscle (P = .008). Ferumoxytol-enhanced MR imaging can depict DVT with a dual-contrast mechanism and show the extent of thrombus.

Magnetic resonance imaging of the pancreas at 3.0 tesla: qualitative and quantitative comparison with 1.5 tesla.

OBJECTIVES: We sought to perform a preliminary comparison of signal-to-noise ratio (SNR) and image quality for magnetic resonance imaging (MRI) of the pancreas at 1.5 and 3 T. MATERIALS AND METHODS: Two imaging cohorts were studied using a T2-weighted, single-shot fast spin-echo pulse sequence and a T1-weighted, fat-suppressed 3D gradient-echo pulse sequence. In the first cohort, 4 subjects were imaged using identical imaging parameters before and after contrast administration at 1.5 and 3.0 T. The SNR was quantified for the pancreas as well as for the liver, spleen, and muscle. In a second cohort of 12 subjects in whom the receiver bandwidth was adjusted for field strength, SNR measurements and qualitative rankings of image quality were performed. RESULTS: In the study cohort using identical imaging parameters at both magnetic field strengths, the mean (SD) ratios of SNR at 3.0 to 1.5 T of the single-shot fast spin-echo images for the pancreas, liver, spleen, and muscle were 1.63 (0.39), 1.82 (0.39), 1.45 (0.18), 2.01 (0.16), respectively. For the precontrast fat-suppressed 3D gradient-echo sequence, the corresponding ratios were 1.28 (0.29), 1.26 (0.30), 1.16 (0.27), and 1.76 (0.45), respectively; for the arterial phase, the corresponding ratios were 2.02 (0.28), 1.60 (0.42), 1.47 (0.26), and 1.94 (0.32), respectively; and for the delayed postcontrast phase, the corresponding ratios were 1.63 (0.51), 2.01 (0.25), 1.66 (0.06), and 2.31 (0.47), respectively. The SNR benefit of 3.0 T was significantly greater on contrast-enhanced as compared with noncontrast T1-weighted 3D gradient-echo images. In the second study cohort, SNR was superior at 3.0 T, although the use of a reduced readout bandwidth at 1.5 T substantially diminished the advantage of the higher field system. With qualitative comparison of images obtained at the 2 magnetic field strengths, the fat-suppressed 3D gradient-echo images obtained at 3.0 T were preferred, whereas the single shot fast spin-echo images obtained at 1.5 T were preferred because of better signal homogeneity. CONCLUSIONS: Our results in a small cohort of volunteers and patients demonstrate a marked improvement in SNR at 3.0 T compared with 1.5 T (by a factor of 2 in some cases) when identical imaging parameters were used. The SNR advantage at 3.0 T is diminished but persists when the receiver bandwidth is adjusted for magnetic field strength. The results suggest that 3.0 T may offer promise for improved body MRI, although further technical development to optimize SNR and improve signal homogeneity will be needed before its full potential can be achieved.

Evaluation of intrarenal oxygenation by BOLD MRI at 3.0 T.

PURPOSE: To examine the benefit of using higher field strengths for BOLD MRI to detect changes in renal medullary oxygenation following pharmacological maneuvers. MATERIALS AND METHODS: Renal BOLD MRI, primarily at 1.5 T, has been shown to be useful for monitoring changes in medullary oxygenation status. We performed the present studies on a 3.0 T scanner using a multiple gradient-echo (mGRE) sequence with a multicoil array to acquire 16 T2*-weighted images within a single breath-hold. Data were obtained before and after administration of furosemide (20 mg iv). RESULTS: The baseline renal R2* (mean +/- SE) at 3.0 T was 37.4+/-1.2 Hz in the medulla, and 21.8 +/- 1.2 Hz in the cortex. The BOLD response to furosemide (DeltaR2*) at 3.0 T was 11.8 +/- 1.1 Hz in the medulla, and 3.0 +/- 0.5 Hz in the cortex. CONCLUSION: Higher magnetic field strength is beneficial for renal BOLD MRI studies. The cortico-medullary contrast on the R2* map was significantly improved at 3.0 T, with no evidence of increased bulk susceptibility artifacts. Baseline R2* and DeltaR2* in the renal medulla at 3.0 T were both significantly higher compared to our previously reported data obtained at 1.5 T.