Original Research. Tadateru Sumi, DDS, 1 Misa Sumi, DDS, PhD, 1 Marc Van Cauteren, PhD, 2 Yasuo Kimura, DDS, PhD, 1 and Takashi Nakamura, DDS, PhD 1 *

Transcription

1 JOURNAL OF MAGNETIC RESONANCE IMAGING 25: (2007) Original Research Parallel Imaging Technique for the External Carotid Artery and its Branches: Comparison of Balanced Turbo Field Echo, Phase Contrast, and Time-of-Flight Sequences Tadateru Sumi, DDS, 1 Misa Sumi, DDS, PhD, 1 Marc Van Cauteren, PhD, 2 Yasuo Kimura, DDS, PhD, 1 and Takashi Nakamura, DDS, PhD 1 * Purpose: To evaluate the parallel imaging technique in the external carotid artery and its branches using 3D balanced turbo field echo (3D btfe), 3D phase-contrast (3D PC), and 3D time-of-flight (3D TOF) MR angiography (MRA) sequences. Materials and Methods: A total of 26 healthy volunteer subjects underwent 3D btfe, 3D PC, and 3D TOF MRA with the parallel imaging sensitivity encoding (SENSE) technique. The obtained images were read in a blinded fashion by three radiologists. Interreader and intersequence statistical analyses were performed to compare the visibility of the arteries. Results: Friedman s ranking test demonstrated that there was no significant difference in visibility between any two pairs of sequences for the external carotid artery and its first branches. However, of the three techniques, 3D PC MRA performed the best for the second-order branches (P 0.01) and for overall visibility of the external carotid artery and its branches (P 0.01). The 3D btfe sequence is superior to 3D TOF; however, an effective means of separating arteries from veins and salivary ducts is needed. Conclusion: The combination of parallel imaging and the 3D PC technique is a promising approach for face and neck MRA. Key Words: parallel imaging; MR angiography; phase-contrast; TOF; balanced TFE; external carotid artery J. Magn. Reson. Imaging 2007;25: Wiley-Liss, Inc. 1 Department of Radiology and Cancer Biology, Nagasaki University School of Dentistry, Nagasaki, Japan. 2 Philips Medical Systems , Kohnan Minato-ku, Tokyo , Japan. *Address reprint requests to: T.N., Department of Radiology and Cancer Biology, Nagasaki University School of Dentistry, Sakamoto, Nagasaki , Japan. taku@nagasaki-u.ac.jp Received April 6, 2006; Accepted November 27, DOI /jmri Published online in Wiley InterScience ( IN CONTRAST TO contrast-enhanced MR angiography (CE-MRA), flow-based MR techniques, such as time-offlight (TOF) and phase-contrast (PC) imaging, have never been used successfully to image the face and neck regions (1). PC imaging provides excellent background suppression and directional flow information. However, the PC technique requires a long image acquisition time, which often leads to poor image quality caused by motion artifacts. TOF imaging is susceptible to saturation effects and is dependent on the direction of flow. These properties of the TOF technique are major disadvantages when this technique is applied to the arteries in the face and neck regions, where the external carotid artery and its branches run along complicated courses and have slow flow rates. The recent development of parallel imaging techniques with phased-array coils has greatly shortened the imaging time in MRA (2). One such technique, sensitivity encoding (SENSE), permits unfolding of reduced field of view (FOV) images into full-fov images with commensurate decreases in imaging time and maintenance of spatial resolution (2,3). These techniques have been successfully applied in CE-MRA and other MRA techniques to target various organs. Thus, we hypothesized that the combined use of the parallel imaging technique and PC imaging might improve the efficacy of MRA in the face and neck regions. In this regard, Gu et al (4) suggested the possibility that an accelerated PC technique could greatly improve the image quality of 3D MRA. Coherent steady-state free precession (SSFP; also referred to as balanced fast field echo (bffe), fast imaging with steady-state free precession (truefisp), or fast imaging employing steady-state acquisition (FIESTA)) has been shown to produce high signals from blood flows due to its high T2/T1 ratio (5). Also, inflowing blood that is not yet in the steady state will produce high signal. By combining fat-suppression sequences, such as spectral presaturation with inversion recovery (SPIR), with SSFP (balanced turbo field echo (btfe)), one can obtain blood signal with better background suppression (6). Furthermore, the short repetition time (TR) of the btfe sequence allows extremely rapid imaging Wiley-Liss, Inc. 1028

2 Parallel Imaging Technique for MRA 1029 Figure 1. Representative frames each obtained with varying Q-flows (10 50 cm/second) in the external carotid (C) and facial (F) arteries in one subject. For this subject we used a Q-flow of 30 cm/second so that the flow signals in both arteries would be detectable and the SNR would be minimal. In this study we attempted to determine the efficacy of three MRA techniques (3D PC, 3D btfe, and 3D TOF) for visualizing the external carotid artery and its branches. MATERIALS AND METHODS Subjects We performed MRI on 26 healthy volunteers (10 women and 16 men, mean age 26 5 years; age range years). The study protocol was approved by the ethics committee of our hospital, and informed consent was obtained from all of the volunteers. MRI MRI was performed on a 1.5T MR imager (Gyroscan Intera 1.5T Master; Philips Medical Systems, Best, The Netherlands) with a 110-mm Synergy Flex S coil (Philips Medical Systems). This small-sized surface coil suffers from inherent signal dropoff away from the coil, which can result in strong image intensity inhomogeneity. We used the Constant Level Appearance (CLEAR) postprocessing technique (Philips Medical Systems) to compensate for this effect. CLEAR uses the premeasured sensitivity profile of the coil elements to calculate the compensation that must be applied on the pixel intensities to obtain even image intensity. Consequently, the CLEAR technique improves image quality and suppresses noise in the images, with resultant increases in the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) (7,8). MR Sequences Each of the volunteers underwent imaging with the 3D PC, 3D btfe, and 3D TOF sequences. The 3D PC (TR/ TE/number of signal acquisitions 20 msec/5.6 msec/1) was performed with a 180-mm FOV, matrix dimension, SENSE factor of 2, flip angle of 15, 1.4-mm slice thickness, and 0.7-mm slice gap. To determine an adequate encoded velocity for the PC MRA of the external carotid artery and its branches, we measured the quantitative flows (Q-flows) of both the facial and external carotid arteries in each of the subjects. Parallel imaging can be used to speed up two-dimensional Q-flow measurements. The combination of Q- flow with echo-planar imaging (EPI) and SENSE decreases the echo train length (ETL) and thus minimizes geometric distortion and signal blurring. Q-flow measurements were performed with the following parameters: FOV 160 mm, scan matrix , acquisition resolution (mm) , reconstruction resolution , T1 FFE sequence, SENSE factor of 2, and TR (msec)/te (msec)/ flip angle ( ) 9.8/6.5/15, 8.9/5.6/15, 8.4/5.2/15, 8.2/4.9/15, 8.1/4.8/15, and 8.0/4.7/15 for 10 cm/ second, 20 cm/second, 30 cm/second, 40 cm/second, 50 cm/second, and 100 cm/second Q-flows, respectively. The slice thickness was 5 mm. A SENSE flex S coil was used. The scan time was 1 minute 16 seconds. This protocol provided a frame rate of 15/second. We used an ECG for retrospective cardiac synchronization. Figure 1 shows examples of real-time Q-flow measurements. In this case we chose an encoded velocity of 30 cm/second. The scan time for PC MRA was 7 minutes 41 seconds for 130 acquired slices. The 3D btfe sequence was combined with an SPIR prepulse (TR/TE/number of signal acquisitions/t2 preparation 4.6 msec/2.3 msec/2/50 msec). The 3D image acquisition technique was performed using a 180-mm FOV, SENSE factor of 2, turbo factor of 192, 90 flip angle, matrix dimension, 1.4-mm slice thickness, and 0.7-mm slice gap. The scan time was three minutes 13 seconds for 130 acquired slices. The 3D TOF (TR/TE/number of signal acquisitions 30 msec/6.9 msec/1) was performed using a 180-mm FOV, SENSE factor of 2, 20 flip angle, matrix dimension, 1.4-mm slice thickness, and 0.7-mm slice gap. A gradient-echo technique with five overlapping slabs was used for 3D TOF MRA. This multislab technique was used to reduce the saturation of the inflowing blood signal. The scan time was four minutes 11 seconds for 130 acquired slices. For all sequences, full maximum intensity projection (MIP) images were obtained from contiguous axial image sources, and a regional saturation (REST) slab (40 mm) was placed cranial to each separate slab to suppress venous flow. MRA Interpretation For each MRA sequence the complete sets of axial source images and MIP images from 25 volunteers were evaluated independently by three radiologists. These readers had various levels of clinical experience ranging from 10 to 15 years. The readers were asked to review the images displayed on the EasyVision workstation (Philips Medical Systems). The readers were allowed free access to all of the axial source images and MIP

3 1030 Sumi et al. and 2) if only the proximal part of or the origin of the vessel was visible, the rating was not visible. Interrupted visibility of the vessel was also rated as not visible. Statistical Analysis Friedman s rank test was used to assess intersequence differences in visibility for the first-order branches, the second-order branches, and the overall (first- and second-order) branches. Fisher s exact probability test was used to assess interreader and intersequence differences in visibility of each of the first- and second-order branches. The interreader variability was assessed by means of the Wilcoxon signed-ranks test. These statistical analyses were all performed with the use of Stat- View software (version 4.0). RESULTS Figure 2. Schematic drawing of the external carotid artery and its branches: 1) common carotid artery, 2) internal carotid artery, 3) external carotid artery, 4) superior thyroid artery, 5) lingual artery, 6) facial artery, 7) occipital artery, 8) posterior auricular artery, 9) maxillary artery, 10) transverse facial artery, 11) zygomaticoorbital artery, 12) medial temporal artery, 13) superficial temporal artery, 14) inferior labial artery, 15) superior labial artery, 16) deep lingual artery, 17) sublingual artery, 18) ascending palatine artery, 19) inferior alveolar artery, 20) masseteric artery, and 21) deep temporal artery. images. Before they reviewed the MRA images, the three readers were trained to identify the normal vascular anatomy on the remaining single set of axial source images and MIP images for 3D PC, 3D btfe, and 3D TOF. To minimize the potential effect of learning after repeated readings, the 25 complete sets of 3D PC, 3D btfe, and 3D TOF images (total of 75 sets) were randomized before they were presented to the three readers. Therefore, the orders of presentation of the different techniques and subjects were random. The readers were not blinded to the type of sequence because the axial MRA images were so distinctive that the blinding procedure would not be effective. The readers were not restricted regarding the available reading time; however, they completed the reading within two weeks on average. We defined the vessels that derive directly from the external carotid artery as the first-order branches, and those that derive from the first-order branches as the second-order branches (Fig. 2). A total of 23 vessels were assessed regarding their visualization using the three types of MRA sequences. The vessels on the left side were assessed first. However, if the obtained images were poor, the right sides were presented to the readers after the images were turned over. The readers were asked to grade the visibility of the vessels according to the following criteria: 1) if the complete course of the vessel was visible, the image was rated as visible, Representative MIP images for 3D PC, 3D btfe, and 3D TOF MRA of the external carotid artery and its branches are shown in Figs The 3D TOF images showed substantial interruptions of the branches of the external carotid artery. The 3D btfe images depicted many veins and salivary ducts, as well as the arteries. The 3D PC sequence provided excellent MIP images of the external carotid artery and its branches, with better background suppression compared to the other two techniques. First we compared the visibility of the external carotid artery and its first- and second-order branches using the three MRA sequences (Table 1). The external carotid artery and most of the first branches were equally and highly visible for the three sequences, while the posterior auricular artery was less visible and the transverse facial artery was rarely detectable for all three sequences. In contrast to the first-order branches, many of the second-order branches were poorly visible (Table 1). Friedman s rank test demonstrated that no significant difference in visibility was evident in the external carotid artery or its first-order branches among the three sequences. However, the second-order branches were best visualized by the 3D PC sequence, followed by the 3D btfe sequence (Table 1). Next we assessed the visibility of each of the first- and second-order branches. In advance, we tested whether any significant variability existed in the visibility by each sequence of the first- and second-order branches among the three readers. The Wilcoxon signed-ranks test revealed no significant difference in visibility between any two groups of the three readers in the three MRA sequences (Table 2). Fisher s exact probability test demonstrated that in comparison with the 3D TOF sequence, the 3D PC sequence better visualized one of the eight first-order branches (the superior thyroid artery) and six of the 14 second-order branches (the submental, superior and inferior labial, angular and deep lingual arteries, and the alar branches of the facial artery) (Table 3).The 3D PC sequence also better visualized one of the second-order branches (the middle meningeal artery) compared to the 3D btfe sequence. The 3D

4 Parallel Imaging Technique for MRA 1031 btfe sequence visualized three of the 14 second-order branches better than the 3D TOF sequence. DISCUSSION We have shown in this paper that the SENSE parallel imaging technique is clinically feasible for nonenhanced MRA of the external carotid artery and its branches. By applying this technique to the 3D PC sequence, we were able to obtain high visibility of the external carotid artery and its first-order branches within eight minutes. We did not observe any significant difference in visibility of the external carotid artery Figure 4. 3D btfe MRA of the external carotid artery and its branches in a 39-year-old man (the same subject shown in Fig. 3). a: Axial MIP image. b: Coronal MIP image. c: Oblique-lateral MIP image. Figure 3. 3D TOF MRA of the external carotid artery and its branches in 39-year-old man. a: Axial MIP image. b: Coronal MIP image. c: Oblique-lateral MIP image. and its first-order branches among the 3D PC, 3D btfe, and 3D TOF sequences (Table 1). However, the second-order branches were significantly better visualized by the 3D PC sequences compared to the other two sequences. The parallel imaging technique enables a reduction in the image acquisition time of 3D PC and accurate Q-flow measurements without breath-holding, and thus greatly improves the image quality of MRA.

5 1032 Sumi et al. Figure 5. 3D PC-MRA of the external carotid artery and its branches in a 39-year-old man (the same subject shown in Figs. 3 and 4). a: Axial MIP image. The inset shows an oblique image of the superior labial artery. b: Coronal MIP image. c: Oblique-lateral MIP image. The recently developed technique of parallel imaging with phased-array coils is becoming widely accepted for MRA applications (2). The major benefit of parallel imaging is that it shortens the acquisition time. Thus, the PC MRA, which otherwise requires lengthy scans, directly benefits from the parallel imaging technique. A 3D PC MRA scan takes four times longer than a regular gradient-echo sequence. Parallel imaging can reduce scan times without increasing artifacts or compromising resolution, and causes minimal reduction of the SNR. For example, the use of a SENSE factor of 2 leads to a decrease in SNR by a factor of about 1/ 2, depending on the distance between the anatomical structures and the coil element. Therefore, the parallel imaging technique may be an ideal adjunct to the lengthy 3D PC sequence. Compared to the TOF sequence, the PC sequence has the advantage of not being dependent on the direction of blood flow. Unlike the major trunk arteries, the branches from the external carotid artery run along complicated courses in the face and neck regions. Therefore, the above-mentioned advantage of the PC technique over TOF enables better visualization of the branch arteries, as evidenced by the present study. In a previous study that compared MRA techniques with respect to identifying the branches of the external carotid artery, the 3D TOF technique performed better than the 3D PC technique (1). These findings are inconsistent with the present findings. In the previous study, however, the differences in slice thickness among the techniques may have significantly contributed to the resolution of each technique. In addition, we speculate that the lengthy image acquisition time used in the previous study (11 minutes 39 seconds for 50 slices vs. seven minutes 41 seconds for 130 slices in the present study) may have greatly contributed to the poor results obtained by the 3D PC technique compared to those obtained by the 3D TOF technique, which required less image acquisition time (five minutes 49 seconds for 135 slices). However, for stenotic vessels, in which flow can be complex, the PC technique may be more difficult to apply because of the longer TEs required. Therefore, the conclusions presented here indicating a preference for the PC technique would appear to be limited to the case of nonstenotic vessels. In TOF MRA, it is preferable to use three-dimensional data acquisition techniques to depict the branches of the external carotid artery, which are tortuous in the face and neck regions. The SENSE technique per se does not reduce saturation effects from a slow and complex arterial course; however, it does allow reduction of the acquisition time and can be used to increase spatial resolution and/or increase the number of slices while maintaining the acquisition time. Improved SNR and anatomic coverage then allow the depiction of small and tortuous arteries, such as the branches of the external carotid artery. In addition to excellent background suppression and minimized saturation effects, the PC technique has an advantage over the TOF method in that it employs variable velocity encoding, which enables imaging of slow and fast flows (9). While the parallel imaging technique is generally used to reduce scan time, and hence improve spatial resolution within the conventional scan time, it can also be beneficial for Q-flow measurements (2). Recently, real-time MR flow techniques have been introduced to save time and to overcome the limitations of conventional PC Q-flow measurements (10). The realtime Q-flow measurement technique, in combination with EPI and SENSE, reduces the ETL and requires

6 Parallel Imaging Technique for MRA 1033 Table 1 Visibility of the External Carotid Artery and Its Branches by 3D PC, 3D TOF, or 3D btfe MRA* Arteries 3D PC 3D TOF 3D btfe A B C Average A B C Average A B C Average Common carotid Internal carotid External carotid First order Superficial temporal Transverse facial Maxillary Posterior auricular Occipital Facial Lingual Superior thyroid Friedman ranking test for firstorder arteries a Second order Middle meningeal Inferior alveolar Anterior deep temporal Posterior deep temporal Masseteric Buccal Ascending palatine Submental Superior labial Inferior labial Alar branch of facial Angular Sublingual Deep lingual Friedman ranking test for second-order arteries a,b Friedman ranking test for overall arteries a,b *Visibility data are expressed as percentages of set MR images where a particular artery was identified by one of the three readers (A, B, and C). a Friedman ranking test was performed to evaluate the difference in visibility among the 3D PC, 3D TOF, and 3D btfe techniques. b Friedman ranking test indicates that the rankings are significant (P 0.01) for the three MRA sequences. Table 2 Interreader Variability in the Detection of the External Carotid Artery and its Branches* MRA sequence Readers (P value) A vs. B B vs. C A vs. C 3D PC D TOF D btfe *Interreader variability was assessed by the Wilcoxon signed-ranks test. only one cardiac cycle, and thus minimizes distortion and blurring errors. The relatively short TRs employed in btfe-based MRA lead to considerably shorter acquisition times compared to conventional segmented k-space gradientecho imaging. Additional specific features of the btfe technique, such as complete refocusing of all available MR signals and a relatively high T2/T1 ratio in blood, allow blood vessels to maintain a dramatically more intense signal over a longer distance compared to conventional TOF imaging (6). To the best of our knowledge, the present study is the first to apply the btfe MRA technique to the head and neck region. Consistent with its known properties, we found that, compared to the 3D TOF technique, 3D btfe MRA offered equivalent visibility for the external carotid artery and its firstorder branches, and better visibility for the secondorder branches. The btfe technique is a powerful tool for visualizing the blood vessels without using any contrast medium. However, it is often difficult to obtain a definite separation between arteries and veins, and even between the blood vessels and salivary gland ducts. Notwithstanding this shortcoming, the application of btfe for visualizing blood vessels may be beneficial for evaluating the extent of tumors (e.g., to locate facial nerves or predict the extension of cancer into neighboring blood vessels) (11,12). A comparison with findings from previous studies that employed 3D PC and 3D TOF sequences (1) shows that the visibility of the carotid artery branches has been greatly improved. However, many branches are still poorly visible even when the 3D PC technique is used (Table 1). Low visi-

7 1034 Sumi et al. Table 3 Comparison in Visibility of Each of the Branches of the External Carotid Artery* Arteries and branches Intersequence differences (P value) PC vs. TOF PC vs. btfe btfe vs. TOF Common carotid Internal carotid External carotid First order Superficial temporal Transverse facial Maxillary Posterior auricular Occipital Facial Lingual Superior thyroid Second order Middle meningeal Inferior alveolar Anterior deep temporal Posterior deep temporal Masseteric Buccal Ascending palatine Submental Superior labial Inferior labial Alar branch of facial Angular Sublingual Deep lingual < < *P values in bold are statistically significant (P 0.05) between the two selected MRA sequences. bility for some of the first-order arteries, such as the transverse facial and posterior auricular arteries, may be due to the orientation of the course of these arteries. The poor visibility observed with the second-order branches may be due in part to the fact that these branches (the inferior alveolar artery, deep temporal artery, ascending palatine artery, and sublingual artery) run through the deep part of the face. It is plausible that the surface coil used in the present study was inadequate for depicting arteries that run in the deep part of the face. To compensate for this as much as possible, we used the CLEAR technique; however, parallel imaging using a head coil suitable for the deep part of the face should be considered as an alternative technology. 2D SENSE appears to be a promising technique for MRA (13); however, it was not available for our MR system when we acquired data for this paper. This technique has advantages over 1D parallel imaging, such as improving the geometry factor with a resultant increase in SNR, and may provide a significant additional reduction in scan time. The practical merits of using this technique await assessment in future studies. In conclusion, while it remains to be determined whether 3D PC MRA coupled with parallel imaging will be able to visualize the external carotid artery and its branches as well as CE-MRA, the parallel imaging technique is a beneficial adjunct for non-ce-mra techniques. Of the three sequences examined in this study, 3D PC yielded the best results. This is the first report to describe the application of btfe MRA to the external carotid artery and its branches. The btfe technique is a powerful tool for visualizing the blood vessels in the head and neck region; however, it remains difficult to distinguish between arteries and veins. ACKNOWLEDGMENT We thank Professor M. Ohki for his suggestions regarding the statistical analysis of our data. REFERENCES 1. van den Berg R, Wasser MNJM, van Gils APG, van der Mey AGL, Hermans J, van Buchem MA. Vascularization of head and neck paragangliomas: comparison of three MR angiographic techniques with digital subtraction angiography. AJNR Am J Neuroradiol 2000:21: Wilson GJ, Hoogeveen RM, Willinek WA, Muthupillai R, Maki JM. Parallel imaging in MR angiography. Top Magn Reson Imaging 2004:15: Golay X, Brown SJ, Itoh R, Melhem ER. Time-resolved contrastenhanced carotid MR angiography using sensitivity encoding (SENSE). AJNR Am J Neuroradiol 2001:22: Gu T, Korosec FR, Block WF, et al. PC VIPR: high-speed 3D phasecontrast method for flow quantification and high-resolution angiography. AJNR Am J Neuroradiol 2005:26: Foo TKF, Ho VB, Marcos HB, Hood MN, Choyke PL. MR angiography using steady-state free precession. Magn Reson Med 2002:48: Coenegrachts KL, Hoogeveen RM, Vaninbroukx JA, et al. Highspatial-resolution 3D balanced turbo field-echo technique for MR angiography of the renal arteries: initial experience. Radiology 2004:231: de Vries M, Nijenhuis RJ, Hoogeveen RM, de Haan MW, van Engelshoven JMA, Leiner T. Contrast-enhanced peripheral MR angiography using SENSE in multiple stations: feasibility study. J Magn Reson Imaging 2005:21: van den Brink JS, Watanabe Y, Kuhl CK, et al. Implications of SENSE MR in routine clinical practice. Eur J Radiol 2003:46: Huston J, Ehman RL. Comparison of time-of-flight and phasecontrast MR neuroangiographic techniques. Radiographics 1993: 13: Hoogeveen R, Leone B, van der Brink J. Real-time quantitative flow using EPI and SENSE. In: Proceedings of the 9th Annual Meeting of ISMRM, Glasgow, Scotland, 2001 (Abstract 114). 11. Sumi M, Van Cauteren M, Takagi Y, Nakamura T. Balanced turbo field echo sequence for MR imaging of parotid gland diseases. AJR Am J Roentgenol 2007:188: El-Hakim H, Mountain R, Carter L, Nilssen EL, Wardrop P, Nimmo M. Anatomic landmarks for locating parotid lesions in relation to the facial nerve: cross sectional radiologic study. J Otolaryngol 2003:32: Hu HH, Madhuranthakam AJ, Kruger DG, Glockner JF, Riederer SJ. Combination of 2D sensitivity encoding and 2D partial Fourier techniques for improved acceleration in 3D contrast-enhanced MR angiography. Magn Reson Med 2006:55:16 22.