Simultaneous Acquisition of MR Angiography and Venography (MRAV)

Transcription

1 Simultaneous Acquisition of MR Angiography and Venography (MRAV) Yiping P. Du 1,2 * and Zhaoyang Jin 3,4 Magnetic Resonance in Medicine 59: (2008) A dual-echo pulse sequence for simultaneous acquisition of MR angiography and venography (MRAV) is developed. Data acquisition of the second echo for susceptibility-weighted imaging based MR venography is added to the conventional three-dimensional (3D) time-of-flight (TOF) MRA pulse sequence. Using this dual-echo acquisition approach, the venography data can be acquired without increasing the repetition time, and, therefore, the scan duration of routine TOF MRA scans is maintained. The feasibility of simultaneous acquisition of MRAV is presented in brain scans at different spatial resolutions. The effect of spatial resolution on vein-to-background contrast is also demonstrated. Venous contrast is improved in high-resolution ( mm 3 ) images compared to that in standardresolution ( mm 3 ) images. This MRAV technique enables the acquisition of MR venography without the need of an extra scan or injection of contrast agent in routine clinical brain exams at 3T. Magn Reson Med 59: , Wiley-Liss, Inc. Key words: susceptibility-weighted imaging; MR venography; MR angiography; pulse sequence; time-of-flight MR susceptibility-weighted imaging (SWI) has recently demonstrated great clinical significance in the diagnosis of several intracranial venous lesions and diseases (1 4). SWI utilizes the relative phase and magnitude change in the venous vasculature introduced by the susceptibility difference between venous blood and parenchyma. The SWI-based MR venography (MRV) technique is capable of depicting venous vasculature in the submillimeter range without using exogenous contrast agent (5). MRV provides unique insight into the pathology and additional diagnostic information on arteriovenous malformation (2,4), brain tumors (6), stroke and hemorrhage (7), multiple sclerosis (8), cavernous and venous angiomas (9), venous sinus thrombosis (9), and traumatic brain injury (10). MRV has also shown to be sensitive in detecting abnormal iron deposition in the iron-related brain diseases, such as Alzheimer s disease (11). 1 Department of Psychiatry, University of Colorado Denver School of Medicine, Denver, Colorado, USA. 2 Department of Radiology, University of Colorado Denver School of Medicine, Denver, Colorado, USA. 3 Department of Biomedical Engineering, Zhejiang University, Hangzhou, China. 4 Institute of Information and Control, Hangzhou Dianzi University, Hangzhou, China. Grant sponsor: National Institutes of Health (NIH); Grant numbers: MH68582, MH070037, MH47476, DA *Correspondence to: Yiping P. Du, Ph.D., Brain Imaging Center, University of Colorado Denver School of Medicine, Mail Stop F-478, PO Box 6508, Aurora, CO Yiping.Du@UCHSC.edu Received 12 October 2007; revised 30 November 2007; accepted 10 January DOI /mrm Published online in Wiley InterScience ( Wiley-Liss, Inc. 954 MR angiography (MRA) based on the time-of-flight (TOF) contrast (12), on the other hand, provides excellent details of arterial vasculature and is routinely used in clinical brain exams (13,14). Intracranial and cervical MRA at 3T has shown increased signal-to-noise ratio (SNR) and improved delineation of arterial lesions compared to that at 1.5T (15). MRA and MRV reveal different neuronal and vascular abnormalities and provide complementary diagnostic assessments of brain diseases. For example, MRA can depict the feeding vessels of a tumor and MRV can identify the draining veins of the tumor (6). It is increasingly desirable to acquire both MRA and MRV in clinical brain exams. Both TOF-based MRA and MRV scans, however, are relatively long, typically ranging from several minutes to 15 minutes. The increased scan time for undergoing both MRA and MRV scans can cause extra patient discomfort and motion, especially when large volume coverage and high resolution are required. The impact on the patient throughput due to the increase of scan time can also adversely affect the adoption of MRV in routine clinical exams. In this study, we present a new technique in which the data acquisition for SWI-MRV is incorporated into the 3D TOF MRA scan: the first echo is used for MRA and the second echo is used for MRV. The resulting venogram has SWI-contrast, not TOF-contrast. This approach of simultaneous acquisition of MR angiography and venography (MRAV) provides additional MRV images without increasing the typical scan time for MRA at 3T. On the other hand, venous contrast (i.e., vein-to-background contrast) in SWI is the result of several effects, including out-of-phase partial volume signal cancellation, magnetic field disturbance outside the veins, and T 2 * shortening and frequency shift of venous blood. Although the exact contrast mechanism is complex, we hypothesize that using a high resolution would improve venous contrast for small veins by reducing partial volume effect. To demonstrate the effect of spatial resolution on venous contrast, in this preliminary study, we selected a standard resolution ( mm 3 ) and a high resolution ( mm 3 ) that are in the range commonly used in MRA and MRV. MATERIALS AND METHODS The data acquisitions of MRA and MRV are similar in several aspects. Both acquisitions use a 3D spoiled gradient-recalled echo (SPGR) pulse sequence and require flow compensation. On the other hand, the difference in the data acquisitions of MRA and MRV is substantial. A short TE, typically at 3 5 msec, is used in MRA to reduce highorder flow artifacts. A relatively long TE is used in MRV to reach out-of-phase partial volume signal cancellation between venous blood and background tissue in the same

2 MR Angiography and Venography 955 voxel, resulting in maximal venous contrast. A cylinder model of venous blood suggests that the venous blood parallel to the main magnetic field is out-of-phase with parenchyma at a TE 28 ms at 3T (16). The active part of the 3D TOF MRA sequence, including excitation, data acquisition, phase rewinding, and gradient spoiling, has a duration of ms. A TR of ms is commonly used in MRA for optimal balance between in-flow effect and background suppression, resulting in a considerable idle time after readout in a TR. At 3T, the idle time in the MRA pulse sequence can be effectively used for the data acquisition of MRV without the need to increase TR. It is therefore logical to investigate the feasibility of acquiring both MRA and MRV datasets in the same pulse sequence. Pulse Sequence The acquisition of a second echo is added to a conventional multiple overlapped thin slab acquisition (14) of a 3D TOF pulse sequence (Fig. 1). By using a fly-back gradient trapezoid, the second readout gradient has the same polarity as the first readout gradient. The fly-back gradient is placed in the middle of two readouts to restore the flow-compensation along the readout direction in the second echo, which can either be a partial or full echo. In this study, a partial echo acquisition for the second echo is selected because it allows a shortened TR for the selected echo time for the second echo (TE2). FIG. 1. Diagram of the dual-echo MRAV pulse sequence. MRA data are acquired at the first echo. MRV data are acquired at the second echo, as indicated by the dashed rectangle. Flow compensation was applied to the readout (G ro ) direction. A fly-back gradient was applied to the readout direction to refocus the second echo. The fly-back gradient was located in the middle of both echoes to restore flow compensation in the second echo. The second echo was acquired with partial echo to either reduce TR or increase the second echo time (TE2). Phase rewinders were applied to the sliceselection and phase-encoding (PE) directions, followed by a gradient spoiler that destroys residual transverse magnetization. Data Acquisition The MRA and MRV data of a healthy subject were acquired using the dual-echo MRAV pulse sequence on a GE 3T scanner (Milwaukee, WI, USA) at standard and high resolutions with a standard birdcage head coil. Standard consent procedure was complied. Both scans have: a rectangular field-of-view (FOV) 20 cm 16 cm, a slice thickness 1.6 mm, and a flip angle 20. A TE ms was selected for near maximal phase cancellation between the veins and parenchyma at 3T in both scans (16). A minimum TR of 32 ms, limited by the readout and spoiling gradient timings, was chosen to minimize the scan duration. A relatively low readout bandwidth of 15.6 khz was selected to ensure sufficient SNR in the MRV data through increasing the readout duration. A total of two slabs with 32 slices per slab were acquired. Flow compensation was applied along the frequency encoding direction to reduce flow artifacts in both echoes. High-order autoshim was applied prior to the scans to improve the field homogeneity. In the MRAV scan with standard resolution, each slab had a matrix size , with an in-plane resolution mm 2, and a voxel volume 0.98mm 3. A 68.7% partial echo was used to reduce the TE of the first echo (TE1 3.1 ms) and maximize TE2 in the second echo. The scan time was 7 min and 13 s. High-resolution MRAV data were acquired with a matrix size at the same rectangular FOV immediately after the standard-resolution acquisition. The in-plane resolution mm 2 and the voxel volume 0.43 mm 3. These voxel dimensions were consistent with the optimal dimensions for MRA at 3T (15). A 66.7% partial echo was used to reduce TE1 to 4.1 ms. The scan time was 10 min and 46 s. Image Reconstruction and Data Analysis Image Reconstruction and postprocessing were performed using MATLAB (TheMathWorks, Inc., Natick, MA, USA). The projection onto convex sets (POCS) (17) algorithm was applied to k-space data along the readout direction (i.e., k x ) with four iterations. Zero-filled interpolation (18) was then applied to k-space data (with both standard and high resolutions) to reconstruct 3D complex images with a matrix size in each slab. The MRA data were obtained by taking the magnitude of the 3D images reconstructed from the first echo. A total of four slices on the top and bottom of each slab were removed to reduce the artifacts introduced by signal wrap-around along the slice direction. Two slabs of the MRA data were then concatenated with four slices overlapped at the boundary. The 3D complex MRV images were reconstructed from the second echo data. A 3D phase mask was constructed from the complex MRV images for each slab to enhance the venous contrast and remove the incidental phase variations due to static magnetic field inhomogeneity effects (3). A rectangular window was applied to the 2D k-space data from each slice, with a filter size of for the MRV with standard resolution and for the MRV with high resolution. These filter sizes were found to yield sufficiently homogeneous phase images while preserving the phase changes in the venous vasculature. The original complex images were divided by the low-pass filtered complex images. The resulting high-pass filtered complex images were used to construct a 3D phase mask by setting the phase values above a threshold of 0 to unity, whereas all the phase values below the threshold and larger than were linearly scaled between zero and unity (3). The mag-

3 956 Du and Jin nitude MRV images were multiplied by the 3D phase mask four times to enhance the visibility of venous vasculature. A total of four slices on the top and bottom of each slab were removed to reduce the wrap-around artifacts. A venogram was generated by performing a minimum-intensity projection (mip) in each slab along the slice direction. RESULTS Figure 2 shows the projections of the standard resolution ( ) data in the top row and high resolution ( ) data in the bottom row. The left column shows the mip of MRV of 32 interpolated slices in the inferior slab with a thickness of 25.6 mm. The middle column shows the mip of MRV of 40 interpolated slices in the superior slab with a thickness of 32.0 mm. The right column shows the maximum-intensity projection (MIP) of MRA of the concatenated slabs. Venous contrast is higher with the high-resolution acquisition compared to that with the standard resolution, as indicated by the black arrows in MRV images in the bottom row. Small arteries are better depicted in the high-resolution MRA, as indicated by the white arrows at the right column. The image intensity profiles along the right/left direction indicated by a black line in the middle column were plotted in Fig. 3. The intensity plot of the standard-resolution MRV is at the top row; the intensity plot of highresolution MRV is at the bottom row. Venous contrast is higher at the high-resolution MRV in vessel segments, as indicated by the short arrows for the increased number and amplitude of downward peaks. FIG. 3. Intensity plots of MRV along the black line shown in the middle column at standard resolution (upper row) and high resolution (bottom row). The short arrows indicate the veins that have higher contrast at high resolution. Figure 4 shows the mip of the 3D phase mask with the standard resolution (left) and high resolution (right) of the superior slab. The phase contrast of small veins is substantially higher at high resolution than that at standard resolution. The profiles of phase mask value along the black lines are shown in Fig. 5. The location of the black line shown in Fig. 4 is the same as that in Fig. 2. In Fig. 5, the veins indicated by the thin arrows have higher contrast at high resolution than those at standard resolution. The short arrows mark the same locations as in Fig. 3. Some of the locations marked with the short arrows overlap with that marked by the thin arrows; and other locations marked FIG. 2. Projections of the standard-resolution ( mm 3 ) data in the top row and high-resolution ( mm 3 ) data in the bottom row. The left column shows the mip of MRV of 32 interpolated slices in the inferior slab with a thickness of 25.6 mm. The middle column shows the mip of MRV of 40 interpolated slices in the superior slab with a thickness of 32.0 mm. The right column shows the MIP of MRA of the concatenated slabs. The veins indicated by the black arrows and the arteries indicated by the white arrows have improved contrast of the high resolution. FIG. 4. mip of the 3D phase mask with the standard resolution (left) and high resolution (right) of the superior slab.

4 MR Angiography and Venography 957 FIG. 5. Profiles of phase mask value along the same black line as shown in Figs. 2 and 4. The veins indicated by the thin arrows have higher contrast at high resolution than that at standard resolution. The short arrows mark the same locations as in Fig. 3. with the short arrows are different than that marked by the thin arrows. This observation suggests that the improved venous contrast at high resolution arises from the improved contrast in both phase and magnitude images. DISCUSSION SNR is still an issue of major concern in SWI-MRV at high spatial resolution even at 3T, although SNR is not a priority concern for TOF or contrast-enhanced based MRV at 3T. The SNR in MRV is lower than that in MRA partly due to the required long TE for MRV. The averaged SNR in the MRV data in the entire imaged brain volume was measured to be about 35% lower than that in the MRA data due to T 2 * decay, consistent with the theoretically estimated signal reduction. Furthermore, venous blood has a lower SNR than stationary tissue due to its negative contrast. Applying multiple multiplications of the phase mask to the original dataset introduces extra noise from the phase mask to the final images. Increasing the readout duration, through reducing the readout bandwidth, can be effective in increasing the SNR of MRV due to the slow flow rate and lack of pulsation in the venous flow. Excluding part of the high-frequency data in k-space during the reconstruction of MRV is another alternative to increasing SNR with the penalty of reduced spatial resolution. The voxel aspect ratio was found to play an important role in the venous contrast of MRV. Experimental observation suggests that the aspect ratio for optimal venous contrast of a vein perpendicular to the main field is R/w/ h 1:1:4, where R is the diameter of the vein, and w and h are the width and height of the voxel dimension (19). For small veins of 0.5 mm in size, an in-plane resolution of 0.5 mm 0.5 mm with a slice thickness of 2 mm will yield the highest venous contrast for transverse acquisitions. This voxel aspect ratio is larger than that typically used in MRA, in which thinner slices are used to preserve the vessel sharpness in projections along nontransverse planes. In our study, the voxel aspect ratio was 1:2.05 in the images with standard resolution and 1:3.07 in the images with high resolution. The improved venous contrast in MRV with high resolution is probably in part due to its increased voxel aspect ratio. The voxel aspect ratio can be altered during image reconstruction. Removing several high frequency views along kz (in the slice direction) would increase the effective slice thickness and, therefore, the voxel aspect ratio in MRV for the enhancement of venous contrast. Using a high voxel aspect ratio, however, will degrade the image quality in sagittal and coronal mip display of the MRA and MRV data. Venous contrast in SWI arises from the susceptibility difference between deoxygenated blood in the vein and surrounding tissue. A vein parallel to the main field has a negative phase inside the vessel. The apparent phase and the degree of out-of-phase signal cancellation in a voxel depend on the partial volume of venous blood in the voxel. A vein perpendicular to the main field has a positive phase inside the vessel and a varying phase, positive or negative, outside the vessel. The apparent phase of a voxel containing the vein, therefore, depends on both the size and location of the voxel (19). For a voxel containing a vein in arbitrary orientation, the apparent phase and the degree of out-ofphase signal cancellation of the voxel also depend on the direction of the vein. Despite the complexity of the contrast mechanism, the preliminary results of two commonly used resolutions (i.e., 0.98 mm 3 and 0.43 mm 3 ) suggest that using a higher spatial resolution results in an improved venous contrast in both phase masks and magnitude images. Using simultaneous acquisition of MRAV eliminates possible misregistration between the arterial and venous vasculatures that otherwise could be induced by interscan patient motion, especially considering the relatively long acquisition time of both scans. The elimination of misregistration can make simultaneous display of MRA and MRV datasets more precise, which could be of great merit for clinicians to identify the exact spatial relationship between the arterial and venous vasculatures at or near lesions. Such exact coregistration between MRA and MRV can also be beneficial in interventional or stereotactic procedures. The MRAV technique can also be applied at 1.5T. A two times longer TE is required at 1.5T to achieve the same level of the susceptibility weighting compared to that at 3T.ATE 56 ms would be needed for blood in veins parallel to the main field to be out-of-phase with parenchyma (16). As such, the optimal TR for MRV is much longer than that for MRA. Adding acquisition for MRV would increase the scan time even with the dual-echo MRAV technique. The increase of TR required for MRV would also affect the vascular contrast in MRA because of the competing effects of the increase in-flow and weakened saturation of the stationary tissue. Careful selection

5 958 Du and Jin of TR is necessary for optimal balance of vascular contrast in the MRA and MRV at 1.5T. Contrast-enhanced MRA and MRV have been used for clinical diagnosis of cerebral vascular diseases. With the injection of contrast agent, high-resolution MRA and MRV with increased vascular contrast can be obtained with a reduced scan time. A technique using the time difference between the arterial phase and venous phase of a single contrast injection has been proposed for simultaneous acquisition of high-resolution intracranial MRA and MRV (20). The administration of gadolinium-based MR contrast agent, however, has recently been found to be linked to the risk of developing nephrogenic systemic fibrosis in patients with impaired renal function (21). This MRAV technique proposed in the present study offers a viable alternative for these patients since it does not require the injection of contrast agent. For simplicity, several techniques commonly used in 3D TOF MRA were not used in this study. These techniques include RF spatial saturation that reduces signal from downward venous flow, tilted optimized nonsaturating excitation (TONE) that makes the upward inflow arterial blood appear more uniform (22), and the magnetization transfer (MT) pulse that saturates signal from stationary tissue. RF spatial saturation is expected to have a minimal effect on the visibility of small veins because of their low flow rate. With the TONE excitation, the venous and background signal would be lower at the inferior end and higher at the superior end of the slab in both MRA and MRV. This signal variation can reduce venous contrast in the original magnitude image and MRV, especially with mip along the slab direction. The signal variation of background tissue along the slab direction, however, can be theoretically estimated with Bloch equations and can be corrected in the MRV data through postprocessing. With MT preparation, the background tissue would have greater signal reduction than the venous blood, resulting in a reduced SNR and venous contrast in MRV. Investigation on the full impact of these techniques to MRV is desirable. In summary, we have developed a technique for simultaneous acquisition of MRA and MRV data in the same scan using a dual-echo approach. Using this technique, MRV acquisition can be added to routine brain exams that include 3D TOF MRA acquisition at 3T without increasing the scan time. Preliminary results have demonstrated the feasibility of the MRAV technique. The effect of spatial resolution on venous contrast in the phase mask and MRV was also demonstrated. Venous contrast was improved at high resolution in both phase and magnitude images. This technique overcomes a substantial hurdle in the scan time management and can potentially accelerate the adoption of MRV in routine clinical brain exams at 3T without using contrast agent. ACKNOWLEDGMENTS We thank Debra Singel for assistance with the scans. We also thank Drs. Dietmar Cordes, and Mark Brown, for their helpful comments. Z.J. is grateful for the mentoring from Dr. Ling Xia at Zhejiang University. Y.P.D. is a recipient of grants from the National Institutes of Health. REFERENCES 1. Reichenbach JR, Venkatesan R, Schillinger DJ, Kido DK, Haacke EM. Small vessels in the human brain: MR venography with deoxyhemoglobin as an intrinsic contrast agent. Radiology 1997;204: Essig M, Reichenbach JR, Schad LR, Schoenberg SO, Debus J, Kaiser WA. High-resolution MR venography of cerebral arteriovenous malformations. Magn Reson Imaging 1999;17: Haacke EM, Xu Y, Cheng YC, Reichenbach JR. Susceptibility weighted imaging (SWI). Magn Reson Med 2004;52: Sehgal V, Delproposto Z, Haacke EM, Tong KA, Wycliffe N, Kido DK, Xu Y, Neelavalli J, Haddar D, Reichenbach JR. Clinical applications of neuroimaging with susceptibility-weighted imaging. J Magn Reson Imaging 2005;22: Reichenbach JR, Essig M, Haacke EM, Lee BC, Przetak C, Kaiser WA, Schad LR. High-resolution venography of the brain using magnetic resonance imaging. MAGMA 1998;6: Sehgal V, Delproposto Z, Haddar D, Haacke EM, Sloan AE, Zamorano LJ, Barger G, Hu J, Xu Y, Prabhakaran KP, Elangovan IR, Neelavalli J, Reichenbach JR. Susceptibility-weighted imaging to visualize blood products and improve tumor contrast in the study of brain masses. J Magn Reson Imaging 2006;24: Wycliffe ND, Choe J, Holshouser B, Oyoyo UE, Haacke EM, Kido DK. Reliability in detection of hemorrhage in acute stroke by a new threedimensional gradient recalled echo susceptibility-weighted imaging technique compared to computed tomography: a retrospective study. J Magn Reson Imaging 2004;20: Tan IL, van Schijndel RA, Pouwels PJ, van Walderveen MA, Reichenbach JR, Manoliu RA, Barkhof F. MR venography of multiple sclerosis. AJNR Am J Neuroradiol 2000;21: Reichenbach JR, Jonetz-Mentzel L, Fitzek C, Haacke EM, Kido DK, Lee BC, Kaiser WA. High-resolution blood oxygen-level dependent MR venography (HRBV): a new technique. Neuroradiology 2001;43: Shen Y, Kou Z, Kreipke CW, Petrov T, Hu J, Haacke EM. In vivo measurement of tissue damage, oxygen saturation changes and blood flow changes after experimental traumatic brain injury in rats using susceptibility weighted imaging. Magn Reson Imaging 2007;25: Haacke EM, Ayaz M, Khan A, Manova ES, Krishnamurthy B, Gollapalli L, Ciulla C, Kim I, Petersen F, Kirsch W. Establishing a baseline phase behavior in magnetic resonance imaging to determine normal vs. abnormal iron content in the brain. J Magn Reson Imaging 2007;26: Wehrli FW, Shimakawa A, Gullberg GT, MacFall JR. Time-of-flight MR flow imaging: selective saturation recovery with gradient refocusing. Radiology 1986;160: Masaryk TJ, Modic MT, Ross JS, Ruggieri PM, Laub GA, Lenz GW, Haacke EM, Selman WR, Wiznitzer M, Harik SI. Intracranial circulation: preliminary clinical results with three-dimensional (volume) MR angiography. Radiology 1989;171: Parker DL, Yuan C, Blatter DD. MR angiography by multiple thin slab 3D acquisition. Magn Reson Med 1991;17: Bernstein MA, Huston J 3rd, Lin C, Gibbs GF, Felmlee JP. High-resolution intracranial and cervical MRA at 3.0T: technical considerations and initial experience. Magn Reson Med 2001;46: Reichenbach JR, Barth M, Haacke EM, Klarhofer M, Kaiser WA, Moser E. High-resolution MR venography at 3.0 Tesla. J Comput Assist Tomogr 2000;24: Haacke EM, Lindskog ED, Lin W. A fast, iterative, partial-fourier technique capable of local phase recovery. J Magn Reson 1991;92: Du YP, Parker DL, Davis WL, Cao G. Reduction of partial-volume artifacts using zero-filled interpolation in three-dimensional MR angiography. J Magn Reson Imaging 1994;4: Xu Y, Haacke EM. The role of voxel aspect ratio in determining apparent vascular phase behavior in susceptibility weighted imaging. Magn Reson Imaging 2006;24: Al-Kwifi O, Shelef I, Farb RI, Stainsby J, Wright GA. High-resolution imaging of the intracranial arterial and venous systems following a single contrast injection. J Magn Reson Imaging 2006;24: Lauenstein TC, Salman K, Morreira R, Tata S, Tudorascu D, Baramidze G, Singh-Parker S, Martin DR. Nephrogenic systemic fibrosis: center case review. J Magn Reson Imaging 2007;26: Atkinson D, Brant-Zawadzki M, Gillan G, Purdy D, Laub G. Improved MR angiography: magnetization transfer suppression with variable flip angle excitation and increased resolution. Radiology 1994;190: