- Email: [email protected]
Visualization of Lenticulostriate Arteries at 3T
Visualization of Lenticulostriate Arteries at 3T: Optimization of Slice-selective Off-resonance Sinc Pulse–prepared TOF-MRA and Its Comparison with Flow-sensitive Black-blood MRA Sachi Okuchi, MD, Tomohisa Okada, MD, PhD, Koji Fujimoto, MD, PhD, Yasutaka Fushimi, MD, PhD, Aki Kido, MD, PhD, Akira Yamamoto, MD, PhD, Mitsunori Kanagaki, MD, PhD, Toshiki Dodo, MD, Taha M. Mehemed, MD, Mitsue Miyazaki, PhD, Xiangzhi Zhou, PhD, Kaori Togashi, MD, PhD Rationale and Objectives: To optimize visualization of lenticulostriate artery (LSA) by time-of-flight (TOF) magnetic resonance angiography (MRA) with slice-selective off-resonance sinc (SORS) saturation transfer contrast pulses and to compare capability of optimal TOF-MRA and flow-sensitive black-blood (FSBB) MRA to visualize the LSA at 3T. Materials and Methods: This study was approved by the local ethics committee, and written informed consent was obtained from all the subjects. TOF-MRA was optimized in 20 subjects by comparing SORS pulses of different flip angles: 0, 400 , and 750 . Numbers of LSAs were counted. The optimal TOF-MRA was compared to FSBB-MRA in 21 subjects. Images were evaluated by the numbers and length of visualized LSAs. Results: LSAs were significantly more visualized in TOF-MRA with SORS pulses of 400 than others (P < .003). When the optimal TOFMRA was compared to FSBB-MRA, the visualization of LSA using FSBB (mean branch numbers 11.1, 95% confidence interval (CI) 10.0–12.1; mean total length 236 mm, 95% CI 210–263 mm) was significantly better than using TOF (4.7, 95% CI 4.1–5.3; 78 mm, 95% CI 67–89 mm) for both numbers and length of the LSA (P < .0001). Conclusions: LSA visualization was best with 400 SORS pulses for TOF-MRA but FSBB-MRA was better than TOF-MRA, which indicates its clinical potential to investigate the LSA on a 3T magnetic resonance imaging. Key Words: Lenticulostriate artery; slice-selective off-resonance sinc pulse; saturation transfer contrast; flow-sensitive black-blood; MR angiography. ªAUR, 2014
I
mpairment of the lenticulostriate artery (LSA) often leads to lacunar infarction and cerebral hemorrhage, which is recognized as ‘small vessels, big problems (1).’ The LSA branches supply blood to the basal ganglia and its vicinity (2,3), and their occlusion results in infarction of these structures (4,5). Recently, LSA branches have successfully been visualized using 7T (6,7) and 3T (8,9) magnetic resonance imaging (MRI) systems with three-dimensional (3D) time-of-flight (TOF) magnetic resonance angiography (MRA). Whereas at 1.5T, a recent study reported that flowsensitive black blood (FSBB) MRA, which decreases blood
Acad Radiol 2014; 21:812–816 From the Department of Diagnostic Radiology, Kyoto University Graduate School of Medicine, 54 Shogoin Kawaharacho, Sakyo-ku, Kyoto 606-8507, Japan (S.O., T.O., K.F., Y.F., A.K., A.Y., M.K., T.D., T.M.M., K.T.) and Toshiba America Research Institute, Vernon Hills, IL (M.M., X.Z.). Received January 14, 2014; accepted March 4, 2014. Address correspondence to: T.O. e-mail: [email protected] ªAUR, 2014 http://dx.doi.org/10.1016/j.acra.2014.03.007
812
flow signal with weak motion-dephasing gradient, performed better than TOF-MRA for visualization of the LSA (10). Moreover, FSBB-MRA at 1.5T could detect differences in the LSA between patients with lacunar infarction and/or hypertension and control subjects (11). However, at 3T, there is no study so far that has compared LSA visualization using TOF-MRA to FSBB-MRA. In TOF-MRA acquisition, the saturation transfer contrast pulse is often used to reduce background signal, but the blood signal is also reduced, although the blood to background contrast is usually increased. To less reduce the blood signal, slice-selective off-resonance sinc (SORS) pulse (12) can be applied, and enhanced visualization of LSA branches at 1.5T has been achieved (13). Therefore, we hypothesized that SORS pulse–prepared TOF-MRA may have high visualization capability of LSA branches comparable to FSBB-MRA. In this study, TOFMRA was firstly investigated for the optimal SORS pulse. Then, comparative study was conducted for visualization of the LSA between the optimized TOF-MRA and FSBBMRA.
Academic Radiology, Vol 21, No 6, June 2014
MRA TO VISUALIZE LSA AT 3T: FSBB VERSUS TOF
MATERIALS AND METHODS
Image Analysis
This study was approved by the local ethics committee, and written informed consent was obtained from all the subjects enrolled. TOF-MRA with SORS pulse was firstly optimized for the flip angle by counting number of visualized LSA branches, then the optimized TOF-MRA was compared to FSBB-MRA. Images were evaluated for numbers and length of visualized LSAs.
In the analysis of TOF-MRA optimization, 3D image volume data were transferred to a commercially available workstation (AZE VirtualPlace Lexus; AZE Ltd., Tokyo, Japan) and the following processing and evaluations were conducted. After reorienting the 3D axial image volumes into coronal (perpendicular to the AC–PC line), each of consecutive five slices was projected by maximum intensity (ie, maximum intensity projection [MIP] of 2 mm thickness). Using these images, LSA branches longer than 5 mm were traced and analyzed (10). When an artery branches within 5 mm from the horizontal segment of the middle cerebral artery (MCA) and the proximal (A1) part of anterior cerebral artery (ACA) origin, each branch was counted and measured separately, because >70% of branches were found to originate from common trunks (3). For the comparison between TOF-MRA and FSBBMRA, TOF-MRA image volumes were reoriented into coronal and the same MIP processing was conducted as described previously. FSBB-MRA image volumes were processed by minimum intensity projection (ie, MinIP of 2 mm thickness). Using these images, LSA branches were traced and measured for length with the same manner described previously. The evaluation was conducted by two radiologists independently (S.O. and T.D. both with 7 years of experience).
Subjects
Twenty volunteers (16 men and 4 women; age range 29–74 years, mean 52 years) were enrolled for the optimization study of TOF-MRA. The subjects had history of smoking (n = 5), hypertension (n = 4), diabetes mellitus (n = 2), and brain infarction (n = 1). Twenty-one volunteers (13 men and 8 women; age range 29–82 years, mean 55 years) were enrolled for the comparative study of the optimized TOF-MRA and FSBB-MRA. The subjects had history of smoking (n = 8), hypertension (n = 7), hyperlipidemia (n = 4), diabetes mellitus (n = 3), cerebral infarction (n = 2), and cerebral hemorrhage (n = 1). Image Acquisition
All scans were taken with a 3T MRI unit (EXCELART Vantage Powered by ATLAS, Toshiba Medical Systems Corporation, Otawara-shi, Japan), equipped with a 32-channel head coil. For optimization, 3D TOF-MRA was acquired with three different flip angles of SORS pulses: 0, 400 , and 750 . The 0 flip angle means the SORS pulse is off; the 400 flip angle was applied before each excitation pulse; the 750 pulse was only applied at the k-space center (1/6 of the total k-space lines) because of ‘‘specific absorption rate’’ (SAR) limitation. After obtaining localizing images of three orthogonal axis, TOF-MRA was scanned in the anterior commissure (AC)– posterior commissure (PC) slice orientation with the following parameters: repetition time (TR)/echo time (TE), 35/6.8 milliseconds; flip angle, 20 ; matrix, 384 384; field of view, 192 192 mm; and one axial 3D slab with 60 slices (0.8 mm slice thickness). FSBB-MRA was scanned with the same spatial resolution to the 3D TOF-MRA, but it was in coronal orientation, because coronal orientation was better compared to axial orientation when resolution was the same (14). Scan parameters of FSBB-MRA were as follows: TR/TE, 35/13 milliseconds; flip angle, 15 ; and b value of the motion-dephasing gradient, 0.3 s/mm2 (14). FSBB-MRA was acquired with the same high resolution to that of TOF-MRA. To improve signal-to-nose ratio, TE was shortened by adopting a smaller b value than that at 1.5T (10). For both TOF-MRA and FSBB-MRA, a parallel imaging factor of 2 was used, and scan time was 7 minutes and 24 seconds. Final image volumes were reconstructed into 0.25 0.25 0.4 mm resolution.
Statistical Analysis
Agreements of independent measurements by two radiologists were evaluated using intraclass correlation coefficients. The numbers of visualized branches per subject were compared pairwise among TOF-MRAs with SORS pulses of 0, 400 ,and 750 using two-tailed paired t test. Both average numbers and total length of visualized LSA branches were compared between FSBB and TOF with two-tailed paired t test. Statistical analyses were conducted using a commercially available software package MedCalc (MedCalc Software, Mariakerke, Belgium). A P value <.05 after Holm correction for multiple comparisons was considered statistically significant.
RESULTS The intraclass correlation coefficients of agreement between the two evaluators in counting LSA branches were 0.88 (95% confidence intervals [CI]: 0.70–0.95), 0.85 (95% CI: 0.62–0.94), and 0.94 (95% CI: 0.86–0.98), respectively for TOF-MRAs with SORS pulse of 0, 400 , and 750 , which was an almost perfect agreement (15). The comparison among intraclass correlation coefficients was not statistically different (P > .05). The average numbers of visualized LSA branches per subject were 3.7 (95% CI: 2.9–4.4), 5.3 (95% CI: 4.5–6.1), and 4.1 (95% CI: 3.2–5.0) for SORS pulses of 0, 400 , and 813
OKUCHI ET AL
Academic Radiology, Vol 21, No 6, June 2014
TABLE 1. Number of Visualized LSA Branches for TOF-MRA 0 Branch numbers
3.7 (2.9–4.4)
400 5.3 (4.5–6.1)*
750 ,y
TABLE 2. Branch Numbers and Length of Visualized LSA Branches for TOF (400 ) and FSBB-MRA
4.1 (3.2–5.0)
LSA, lenticulostriate artery; MRA, magnetic resonance angiography; TOF, time-of-flight. The numbers in the parentheses are 95% confidence intervals. *P = .0001 compared with 0 . y P = .0025 compared with 750 .
TOF FSBB P value
Branch Numbers
Length (mm)
4.7 (4.1–5.3) 11.1 (10.0–12.1) <.0001
78 (67–89) 236 (210–263) <.0001
FSBB, flow-sensitive black blood; LSA, lenticulostriate artery; MRA, magnetic resonance angiography; TOF, time-of-flight. The numbers in the parentheses are 95% confidence intervals.
for counting LSA branches were 0.78 (95% CI: 0.45–0.91) and 0.89 (95% CI: 0.73–0.96) and those for measuring LSA branch length were 0.68 (95% CI: 0.22–0.87) and 0.87 (95% CI: 0.67–0.95) for TOF-MRA and FSBB-MRA, which was substantial to almost perfect agreement (15). The comparison of intraclass correlation coefficient was not statistically different (P = .25 and 0.15, respectively for numbers and length) between the two MRAs. Average numbers of visualized LSA branches per subject were 4.7 (95% CI: 4.1–5.3) and 11.1 (95% CI: 10.0–12.1), and averages of total length of visualized LSA branches were 78 mm (95% CI: 67–89 mm) and 236 mm (95% CI: 210–263 mm), respectively for TOF-MRA and FSBBMRA. In both comparisons, the difference was highly significant (P < .0001; Table 2 and Fig. 2).
DISCUSSION
Figure 1. Representative images of LSA branches (white arrows) in TOF-MRA with SORS pulses of (a) 0 , (b) 400 , and (c) 750 . The LSA branches in TOF-MRA with SORS pulse of 400 were visualized better than those of 0 and 750 . LSA, lenticulostriate artery; MRA, magnetic resonance angiography; SORS, slice-selective off-resonance sinc; TOF, time-of-flight.
750 , respectively. The average LSA branch number with SORS pulse of 400 was statistically greater than those of 0 and 750 (P = .0001 and 0.0025, respectively; Table 1 and Fig. 1). The difference between 0 and 750 was not statistically significant. In the comparison between the optimal TOF-MRA (ie, with SORS pulse of 400 ) and FSBB-MRA, intraclass correlation coefficients of agreement between two evaluators 814
The saturation transfer contrast method has proven to be a powerful method to increase contrast in MRI and MRA, using relaxation differences in tissues (16–18). It was first used in imaging using continuous wave off-resonance irradiation (16). The saturation pulse is spatially nonselective in general and reduces the signal of brain background tissues as well as that of inflowing blood. However, the SORS pulse suppresses brain background signal while maintaining signal of inflowing blood (12). As was expected, the LSA branches were visualized better in TOF-MRA with the SORS pulse than without it (ie, 0 ). The LSA branches were visualized better in TOFMRA with 400 SORS pulse than with 750 pulse, probably because the 400 SORS pulse was placed in front of each excitation, whereas the 750 SORS pulse was placed at the k-space center only. That is the tradeoff between higher SORS pulse saturation power and the SAR limit. In the following comparison study, the average number of visualized LSA branches was more than doubled and the total length of LSA visualization was tripled by FSBB-MRA than the optimized TOF-MRA. The reason is because TOFMRA is a blood inflow technique (19); for the very slow blood flow in LSA branches, the blood signal is more likely to be saturated. For a black-blood scan, the fast-spin echo sequence was initially used (20,21); however, very slow or recirculating vessels were difficult to visualize without incorporating inversion recovery preparation pulses (22,23).
Academic Radiology, Vol 21, No 6, June 2014
Figure 2. Representative images of LSA branches (white arrows) in (a) TOF-MRA (400 ) and (b) FSBB-MRA. LSA branches were visualized better with FSBB than TOF. FSBB, flow-sensitive black blood; LSA, lenticulostriate artery; MRA, magnetic resonance angiography; TOF, time-of-flight.
Weighting on susceptibility contrast has also been used (24,25), but it is more suitable for small veins rather than arteries. FSBB is an alternative black-blood method, where the flow-sensitive gradient can dephase blood signal with a wide range of flow velocity with an appropriate b value (26). Such a difference in visualization mechanism has resulted in the higher visualizing capability of FSBB-MRA. There is an interesting variant of FSBB-MRA, which is hybrid of opposite-contrast MRA. It combines TOF and FSBB methods by subtracting the latter from the former images (27), whose evaluation is yet to be conducted. A recent study reported that FSBB-MRA was better than TOF-MRA at 1.5T for visualization of the LSA (10). We had similar findings to this study at 3T. Using FSBB-MRA, the average numbers of visualized LSA branches per subject were 6.9 at 1.5T (10) compared to 11.1 at 3T. The numbers of LSA branches were reported to range from 2 to 12 (mean 7.1) on one side of hemisphere in autopsied brains (3), meaning 14.2 LSA branches in total on average. FSBB-MRA at 3T has some more to be improved. Detailed visualization of LSA branches is clinically very important. The relationship between decreased LSA visualization and hypertension or infarction at the basal ganglia and/or its vicinity has been well demonstrated (8,11,28–30). If some morphologic changes in the LSA, such as hypovisualization, were detected in asymptomatic patients with hypertension or
MRA TO VISUALIZE LSA AT 3T: FSBB VERSUS TOF
diabetes mellitus, it would motivate patients and physicians for better disease control and may allow more rigorous therapeutic interventions to prevent symptomatic events. Based on the findings in this study, FSBB-MRA can be a good method for visualizing the LSA in clinical setting, although further studies are required to clarify predictability of FSBB findings for future morbidity. Another advantage of FSBB-MRA is visualization capability of the microanatomy around the proximal MCA and its branches, which is very important for planning an aneurysm surgery. It may help to avoid clipping or blood flow disturbance of LSA or other small branches (31,32). Compared to FSBB-MRA, TOF-MRA is more sensitive to inflow speed, which is disadvantageous for the LSA visualization. However, this effect itself may have clinically important information. The spatial resolution of TOF-MRA images in this study was larger than most of the LSA branches. Arterial signal inside of an imaging voxel is averaged with the background and obscured. A smaller voxel improves visualization of the LSA, but the scan time is increased, which is usually not acceptable as a clinical scan. Recently, compressed sensing (CS) has been brought into MRI. CS reconstructs images from undersampled data using iterative steps. This method may improve visualization of the LSA, but the background signal has to be lower than those of LSA branches (33). Reduction of the background signal by the SORS pulse would also contribute to that purpose. There are a few limitations in this study. First, there are several structures visualized as black on FSBB images. Perivascular space or cerebrospinal fluid (CSF) might be detected as the same signal void. In aged subjects, the perivascular space is frequently dilated, but the CSF signal in the space is usually higher than the dephased arterial blood signal of the LSA branch. Calcification, small hemorrhage, and iron deposit are also visualized as a signal void, but these can be identified as spotty, nodular, or even mass-like lesions and may not exhibit the ‘‘string-like’’ shape of LSA branches. Veins can also be visualized as a signal void if the blood flow velocity is close to that in LSA. However in this study, vessels arising from the MCA and the part A1 of ACA were included. Second, some volunteers had history of disorders, which may affect the visualization of LSAs, but the condition was the same for both TOF-MRA and FSBB-MRA. In conclusion, the LSA visualization using TOF-MRA was best with SORS pulse of 400 flip angle than with those of 0 and 750 . However, the optimized TOF-MRA with SORS pulse is not comparable to FSBB-MRA for the visualization of LSA. This indicates that FSBB-MRA would be a better choice at 3T for visualization of the LSA.
ACKNOWLEDGMENTS We express our sincere gratitude to Mrs. Kyoko Takakura, RT and Mr. Hajime Sagawa, RT at Department of Radiology, Kyoto University Hospital and Mr. Naotaka Sakashita at 815
OKUCHI ET AL
Toshiba Medical Systems, Ohtawara-shi, Japan for their help of this study. Grants: This work is funded by a sponsored research program ‘‘Research for Improvement of MR Visualization (No. 150100700014)’’ of Toshiba Medical Systems, Japan, and Grant-in-Aid for Scientific Research on Innovative Areas ‘‘Initiative for High-Dimensional Data-Driven Science through Deepening of Sparse Modeling (No. 4503)’’ of The Ministry of Education, Culture, Sports, Science and Technology, Japan provided to K.T. The former grant contributed to the acquisition of volunteer images of TOF-MRA and FSBB-MRA. The latter grant contributed to analysis of the acquired data. REFERENCES 1. Greenberg SM. Small vessels, big problems. N Engl J Med 2006; 354(14): 1451–1453. 2. Marinkovic SV, Milisavljevic MM, Kovacevic MS, et al. Perforating branches of the middle cerebral artery. Microanatomy and clinical significance of their intracerebral segments. Stroke 1985; 16(6): 1022–1029. 3. Marinkovic S, Gibo H, Milisavljevic M, et al. Anatomic and clinical correlations of the lenticulostriate arteries. Clin Anat 2001; 14(3):190–195. 4. Feekes JA, Hsu SW, Chaloupka JC, et al. Tertiary microvascular territories define lacunar infarcts in the basal ganglia. Ann Neurol 2005; 58(1):18–30. 5. Feekes JA, Cassell MD. The vascular supply of the functional compartments of the human striatum. Brain 2006; 129(Pt 8):2189–2201. 6. Cho ZH, Kang CK, Han JY, et al. Observation of the lenticulostriate arteries in the human brain in vivo using 7.0T MR angiography. Stroke 2008; 39(5): 1604–1606. 7. Kang CK, Park CW, Han JY, et al. Imaging and analysis of lenticulostriate arteries using 7.0-Tesla magnetic resonance angiography. Magn Reson Med 2009; 61(1):136–144. 8. Chen YC, Li MH, Li YH, et al. Analysis of correlation between the number of lenticulostriate arteries and hypertension based on high-resolution MR angiography findings. AJNR Am J Neuroradiol 2011; 32(10):1899–1903. 9. Akashi T, Taoka T, Ochi T, et al. Branching pattern of lenticulostriate arteries observed by MR angiography at 3.0 T. Jpn J Radiol 2012; 30(4): 331–335. 10. Gotoh K, Okada T, Miki Y, et al. Visualization of the lenticulostriate artery with flow-sensitive black-blood acquisition in comparison with time-offlight MR angiography. J Magn Reson Imaging 2009; 29(1):65–69. 11. Okuchi S, Okada T, Ihara M, et al. Visualization of lenticulostriate arteries by flow-sensitive black-blood MR angiography on a 1.5T MRI system: a comparative study between subjects with and without stroke. AJNR Am J Neuroradiol 2013; 34(4):780–784. 12. Miyazaki M, Kojima F, Ichinose N, et al. A novel saturation transfer contrast method for 3D time-of-flight magnetic resonance angiography: a sliceselective off-resonance sinc pulse (SORS) technique. Magn Reson Med 1994; 32(1):52–59.
816
Academic Radiology, Vol 21, No 6, June 2014
13. Admiraal-Behloul1 F, Blink1 E, Zhang1 B, et al. High resolution time-of flight MRA using slice selective saturation transfer contrast and water excitation technique for the visualization of the lenticulostriate arteries at 1.5T. Proc Intl Soc Mag Reson Med 2011; 19:363. 14. Okada T, Miyazaki M, Kido A, Visualization of the lenticulostriate arteries at 3T: a comparative study of TOF-MRA with MTC and FSBB-MRA. Proceedings of the 17th Annual Scientific Meeting of KSMRM. 2012:389. 15. Kundel HL, Polansky M. Measurement of observer agreement. Radiology 2003; 228(2):303–308. 16. Wolff SD, Balaban RS. Magnetization transfer contrast (MTC) and tissue water proton relaxation in vivo. Magn Reson Med 1989; 10(1):135–144. 17. Koenig SH. Cholesterol of myelin is the determinant of gray-white contrast in MRI of brain. Magn Reson Med 1991; 20(2):285–291. 18. Pike GB, Hu BS, Glover GH, et al. Magnetization transfer time-of-flight magnetic resonance angiography. Magn Reson Med 1992; 25(2):372–379. 19. Bradley WG, Jr, Waluch V. Blood flow: magnetic resonance imaging. Radiology 1985; 154(2):443–450. 20. Edelman RR, Mattle HP, Wallner B, et al. Extracranial carotid arteries: evaluation with ‘‘black blood’’ MR angiography. Radiology 1990; 177(1):45–50. 21. Alexander AL, Buswell HR, Sun Y, et al. Intracranial black-blood MR angiography with high-resolution 3D fast spin echo. Magn Reson Med 1998; 40(2):298–310. 22. Mayo JR, Culham JA, MacKay AL, et al. Blood MR signal suppression by preexcitation with inverting pulses. Radiology 1989; 173(1):269–271. 23. Parker DL, Goodrich KC, Masiker M, et al. Improved efficiency in doubleinversion fast spin-echo imaging. Magn Reson Med 2002; 47(5): 1017–1021. 24. Reichenbach JR, Barth M, Haacke EM, et al. High-resolution MR venography at 3.0 Tesla. J Comput Assist Tomogr 2000; 24(6):949–957. 25. Haacke EM, Xu Y, Cheng YC, et al. Susceptibility weighted imaging (SWI). Magn Reson Med 2004; 52(3):612–618. 26. Le Bihan D, Breton E, Lallemand D, et al. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988; 168(2): 497–505. 27. Kimura T, Ikedo M, Takemoto S. Hybrid of opposite-contrast MR angiography (HOP-MRA) combining time-of-flight and flow-sensitive black-blood contrasts. Magn Reson Med 2009; 62(2):450–458. 28. Kang CK, Park CA, Lee H, et al. Hypertension correlates with lenticulostriate arteries visualized by 7T magnetic resonance angiography. Hypertension 2009; 54(5):1050–1056. 29. Kang CK, Park CA, Park CW, et al. Lenticulostriate arteries in chronic stroke patients visualised by 7 T magnetic resonance angiography. Int J Stroke 2010; 5(5):374–380. 30. Gotoh K, Okada T, Satogami N, et al. Evaluation of CT angiography for visualisation of the lenticulostriate artery: difference between normotensive and hypertensive patients. Br J Radiol 2012; 85(1019):e1004–e1008. 31. Sasaki T, Kodama N, Matsumoto M, et al. Blood flow disturbance in perforating arteries attributable to aneurysm surgery. J Neurosurg 2007; 107(1): 60–67. 32. Park DH, Kang SH, Lee JB, et al. Angiographic features, surgical management and outcomes of proximal middle cerebral artery aneurysms. Clin Neurol Neurosurg 2008; 110(6):544–551. 33. Lustig M, Donoho D, Pauly JM. Sparse MRI: the application of compressed sensing for rapid MR imaging. Magn Reson Med 2007; 58(6): 1182–1195.
I
mpairment of the lenticulostriate artery (LSA) often leads to lacunar infarction and cerebral hemorrhage, which is recognized as ‘small vessels, big problems (1).’ The LSA branches supply blood to the basal ganglia and its vicinity (2,3), and their occlusion results in infarction of these structures (4,5). Recently, LSA branches have successfully been visualized using 7T (6,7) and 3T (8,9) magnetic resonance imaging (MRI) systems with three-dimensional (3D) time-of-flight (TOF) magnetic resonance angiography (MRA). Whereas at 1.5T, a recent study reported that flowsensitive black blood (FSBB) MRA, which decreases blood
Acad Radiol 2014; 21:812–816 From the Department of Diagnostic Radiology, Kyoto University Graduate School of Medicine, 54 Shogoin Kawaharacho, Sakyo-ku, Kyoto 606-8507, Japan (S.O., T.O., K.F., Y.F., A.K., A.Y., M.K., T.D., T.M.M., K.T.) and Toshiba America Research Institute, Vernon Hills, IL (M.M., X.Z.). Received January 14, 2014; accepted March 4, 2014. Address correspondence to: T.O. e-mail: [email protected] ªAUR, 2014 http://dx.doi.org/10.1016/j.acra.2014.03.007
812
flow signal with weak motion-dephasing gradient, performed better than TOF-MRA for visualization of the LSA (10). Moreover, FSBB-MRA at 1.5T could detect differences in the LSA between patients with lacunar infarction and/or hypertension and control subjects (11). However, at 3T, there is no study so far that has compared LSA visualization using TOF-MRA to FSBB-MRA. In TOF-MRA acquisition, the saturation transfer contrast pulse is often used to reduce background signal, but the blood signal is also reduced, although the blood to background contrast is usually increased. To less reduce the blood signal, slice-selective off-resonance sinc (SORS) pulse (12) can be applied, and enhanced visualization of LSA branches at 1.5T has been achieved (13). Therefore, we hypothesized that SORS pulse–prepared TOF-MRA may have high visualization capability of LSA branches comparable to FSBB-MRA. In this study, TOFMRA was firstly investigated for the optimal SORS pulse. Then, comparative study was conducted for visualization of the LSA between the optimized TOF-MRA and FSBBMRA.
Academic Radiology, Vol 21, No 6, June 2014
MRA TO VISUALIZE LSA AT 3T: FSBB VERSUS TOF
MATERIALS AND METHODS
Image Analysis
This study was approved by the local ethics committee, and written informed consent was obtained from all the subjects enrolled. TOF-MRA with SORS pulse was firstly optimized for the flip angle by counting number of visualized LSA branches, then the optimized TOF-MRA was compared to FSBB-MRA. Images were evaluated for numbers and length of visualized LSAs.
In the analysis of TOF-MRA optimization, 3D image volume data were transferred to a commercially available workstation (AZE VirtualPlace Lexus; AZE Ltd., Tokyo, Japan) and the following processing and evaluations were conducted. After reorienting the 3D axial image volumes into coronal (perpendicular to the AC–PC line), each of consecutive five slices was projected by maximum intensity (ie, maximum intensity projection [MIP] of 2 mm thickness). Using these images, LSA branches longer than 5 mm were traced and analyzed (10). When an artery branches within 5 mm from the horizontal segment of the middle cerebral artery (MCA) and the proximal (A1) part of anterior cerebral artery (ACA) origin, each branch was counted and measured separately, because >70% of branches were found to originate from common trunks (3). For the comparison between TOF-MRA and FSBBMRA, TOF-MRA image volumes were reoriented into coronal and the same MIP processing was conducted as described previously. FSBB-MRA image volumes were processed by minimum intensity projection (ie, MinIP of 2 mm thickness). Using these images, LSA branches were traced and measured for length with the same manner described previously. The evaluation was conducted by two radiologists independently (S.O. and T.D. both with 7 years of experience).
Subjects
Twenty volunteers (16 men and 4 women; age range 29–74 years, mean 52 years) were enrolled for the optimization study of TOF-MRA. The subjects had history of smoking (n = 5), hypertension (n = 4), diabetes mellitus (n = 2), and brain infarction (n = 1). Twenty-one volunteers (13 men and 8 women; age range 29–82 years, mean 55 years) were enrolled for the comparative study of the optimized TOF-MRA and FSBB-MRA. The subjects had history of smoking (n = 8), hypertension (n = 7), hyperlipidemia (n = 4), diabetes mellitus (n = 3), cerebral infarction (n = 2), and cerebral hemorrhage (n = 1). Image Acquisition
All scans were taken with a 3T MRI unit (EXCELART Vantage Powered by ATLAS, Toshiba Medical Systems Corporation, Otawara-shi, Japan), equipped with a 32-channel head coil. For optimization, 3D TOF-MRA was acquired with three different flip angles of SORS pulses: 0, 400 , and 750 . The 0 flip angle means the SORS pulse is off; the 400 flip angle was applied before each excitation pulse; the 750 pulse was only applied at the k-space center (1/6 of the total k-space lines) because of ‘‘specific absorption rate’’ (SAR) limitation. After obtaining localizing images of three orthogonal axis, TOF-MRA was scanned in the anterior commissure (AC)– posterior commissure (PC) slice orientation with the following parameters: repetition time (TR)/echo time (TE), 35/6.8 milliseconds; flip angle, 20 ; matrix, 384 384; field of view, 192 192 mm; and one axial 3D slab with 60 slices (0.8 mm slice thickness). FSBB-MRA was scanned with the same spatial resolution to the 3D TOF-MRA, but it was in coronal orientation, because coronal orientation was better compared to axial orientation when resolution was the same (14). Scan parameters of FSBB-MRA were as follows: TR/TE, 35/13 milliseconds; flip angle, 15 ; and b value of the motion-dephasing gradient, 0.3 s/mm2 (14). FSBB-MRA was acquired with the same high resolution to that of TOF-MRA. To improve signal-to-nose ratio, TE was shortened by adopting a smaller b value than that at 1.5T (10). For both TOF-MRA and FSBB-MRA, a parallel imaging factor of 2 was used, and scan time was 7 minutes and 24 seconds. Final image volumes were reconstructed into 0.25 0.25 0.4 mm resolution.
Statistical Analysis
Agreements of independent measurements by two radiologists were evaluated using intraclass correlation coefficients. The numbers of visualized branches per subject were compared pairwise among TOF-MRAs with SORS pulses of 0, 400 ,and 750 using two-tailed paired t test. Both average numbers and total length of visualized LSA branches were compared between FSBB and TOF with two-tailed paired t test. Statistical analyses were conducted using a commercially available software package MedCalc (MedCalc Software, Mariakerke, Belgium). A P value <.05 after Holm correction for multiple comparisons was considered statistically significant.
RESULTS The intraclass correlation coefficients of agreement between the two evaluators in counting LSA branches were 0.88 (95% confidence intervals [CI]: 0.70–0.95), 0.85 (95% CI: 0.62–0.94), and 0.94 (95% CI: 0.86–0.98), respectively for TOF-MRAs with SORS pulse of 0, 400 , and 750 , which was an almost perfect agreement (15). The comparison among intraclass correlation coefficients was not statistically different (P > .05). The average numbers of visualized LSA branches per subject were 3.7 (95% CI: 2.9–4.4), 5.3 (95% CI: 4.5–6.1), and 4.1 (95% CI: 3.2–5.0) for SORS pulses of 0, 400 , and 813
OKUCHI ET AL
Academic Radiology, Vol 21, No 6, June 2014
TABLE 1. Number of Visualized LSA Branches for TOF-MRA 0 Branch numbers
3.7 (2.9–4.4)
400 5.3 (4.5–6.1)*
750 ,y
TABLE 2. Branch Numbers and Length of Visualized LSA Branches for TOF (400 ) and FSBB-MRA
4.1 (3.2–5.0)
LSA, lenticulostriate artery; MRA, magnetic resonance angiography; TOF, time-of-flight. The numbers in the parentheses are 95% confidence intervals. *P = .0001 compared with 0 . y P = .0025 compared with 750 .
TOF FSBB P value
Branch Numbers
Length (mm)
4.7 (4.1–5.3) 11.1 (10.0–12.1) <.0001
78 (67–89) 236 (210–263) <.0001
FSBB, flow-sensitive black blood; LSA, lenticulostriate artery; MRA, magnetic resonance angiography; TOF, time-of-flight. The numbers in the parentheses are 95% confidence intervals.
for counting LSA branches were 0.78 (95% CI: 0.45–0.91) and 0.89 (95% CI: 0.73–0.96) and those for measuring LSA branch length were 0.68 (95% CI: 0.22–0.87) and 0.87 (95% CI: 0.67–0.95) for TOF-MRA and FSBB-MRA, which was substantial to almost perfect agreement (15). The comparison of intraclass correlation coefficient was not statistically different (P = .25 and 0.15, respectively for numbers and length) between the two MRAs. Average numbers of visualized LSA branches per subject were 4.7 (95% CI: 4.1–5.3) and 11.1 (95% CI: 10.0–12.1), and averages of total length of visualized LSA branches were 78 mm (95% CI: 67–89 mm) and 236 mm (95% CI: 210–263 mm), respectively for TOF-MRA and FSBBMRA. In both comparisons, the difference was highly significant (P < .0001; Table 2 and Fig. 2).
DISCUSSION
Figure 1. Representative images of LSA branches (white arrows) in TOF-MRA with SORS pulses of (a) 0 , (b) 400 , and (c) 750 . The LSA branches in TOF-MRA with SORS pulse of 400 were visualized better than those of 0 and 750 . LSA, lenticulostriate artery; MRA, magnetic resonance angiography; SORS, slice-selective off-resonance sinc; TOF, time-of-flight.
750 , respectively. The average LSA branch number with SORS pulse of 400 was statistically greater than those of 0 and 750 (P = .0001 and 0.0025, respectively; Table 1 and Fig. 1). The difference between 0 and 750 was not statistically significant. In the comparison between the optimal TOF-MRA (ie, with SORS pulse of 400 ) and FSBB-MRA, intraclass correlation coefficients of agreement between two evaluators 814
The saturation transfer contrast method has proven to be a powerful method to increase contrast in MRI and MRA, using relaxation differences in tissues (16–18). It was first used in imaging using continuous wave off-resonance irradiation (16). The saturation pulse is spatially nonselective in general and reduces the signal of brain background tissues as well as that of inflowing blood. However, the SORS pulse suppresses brain background signal while maintaining signal of inflowing blood (12). As was expected, the LSA branches were visualized better in TOF-MRA with the SORS pulse than without it (ie, 0 ). The LSA branches were visualized better in TOFMRA with 400 SORS pulse than with 750 pulse, probably because the 400 SORS pulse was placed in front of each excitation, whereas the 750 SORS pulse was placed at the k-space center only. That is the tradeoff between higher SORS pulse saturation power and the SAR limit. In the following comparison study, the average number of visualized LSA branches was more than doubled and the total length of LSA visualization was tripled by FSBB-MRA than the optimized TOF-MRA. The reason is because TOFMRA is a blood inflow technique (19); for the very slow blood flow in LSA branches, the blood signal is more likely to be saturated. For a black-blood scan, the fast-spin echo sequence was initially used (20,21); however, very slow or recirculating vessels were difficult to visualize without incorporating inversion recovery preparation pulses (22,23).
Academic Radiology, Vol 21, No 6, June 2014
Figure 2. Representative images of LSA branches (white arrows) in (a) TOF-MRA (400 ) and (b) FSBB-MRA. LSA branches were visualized better with FSBB than TOF. FSBB, flow-sensitive black blood; LSA, lenticulostriate artery; MRA, magnetic resonance angiography; TOF, time-of-flight.
Weighting on susceptibility contrast has also been used (24,25), but it is more suitable for small veins rather than arteries. FSBB is an alternative black-blood method, where the flow-sensitive gradient can dephase blood signal with a wide range of flow velocity with an appropriate b value (26). Such a difference in visualization mechanism has resulted in the higher visualizing capability of FSBB-MRA. There is an interesting variant of FSBB-MRA, which is hybrid of opposite-contrast MRA. It combines TOF and FSBB methods by subtracting the latter from the former images (27), whose evaluation is yet to be conducted. A recent study reported that FSBB-MRA was better than TOF-MRA at 1.5T for visualization of the LSA (10). We had similar findings to this study at 3T. Using FSBB-MRA, the average numbers of visualized LSA branches per subject were 6.9 at 1.5T (10) compared to 11.1 at 3T. The numbers of LSA branches were reported to range from 2 to 12 (mean 7.1) on one side of hemisphere in autopsied brains (3), meaning 14.2 LSA branches in total on average. FSBB-MRA at 3T has some more to be improved. Detailed visualization of LSA branches is clinically very important. The relationship between decreased LSA visualization and hypertension or infarction at the basal ganglia and/or its vicinity has been well demonstrated (8,11,28–30). If some morphologic changes in the LSA, such as hypovisualization, were detected in asymptomatic patients with hypertension or
MRA TO VISUALIZE LSA AT 3T: FSBB VERSUS TOF
diabetes mellitus, it would motivate patients and physicians for better disease control and may allow more rigorous therapeutic interventions to prevent symptomatic events. Based on the findings in this study, FSBB-MRA can be a good method for visualizing the LSA in clinical setting, although further studies are required to clarify predictability of FSBB findings for future morbidity. Another advantage of FSBB-MRA is visualization capability of the microanatomy around the proximal MCA and its branches, which is very important for planning an aneurysm surgery. It may help to avoid clipping or blood flow disturbance of LSA or other small branches (31,32). Compared to FSBB-MRA, TOF-MRA is more sensitive to inflow speed, which is disadvantageous for the LSA visualization. However, this effect itself may have clinically important information. The spatial resolution of TOF-MRA images in this study was larger than most of the LSA branches. Arterial signal inside of an imaging voxel is averaged with the background and obscured. A smaller voxel improves visualization of the LSA, but the scan time is increased, which is usually not acceptable as a clinical scan. Recently, compressed sensing (CS) has been brought into MRI. CS reconstructs images from undersampled data using iterative steps. This method may improve visualization of the LSA, but the background signal has to be lower than those of LSA branches (33). Reduction of the background signal by the SORS pulse would also contribute to that purpose. There are a few limitations in this study. First, there are several structures visualized as black on FSBB images. Perivascular space or cerebrospinal fluid (CSF) might be detected as the same signal void. In aged subjects, the perivascular space is frequently dilated, but the CSF signal in the space is usually higher than the dephased arterial blood signal of the LSA branch. Calcification, small hemorrhage, and iron deposit are also visualized as a signal void, but these can be identified as spotty, nodular, or even mass-like lesions and may not exhibit the ‘‘string-like’’ shape of LSA branches. Veins can also be visualized as a signal void if the blood flow velocity is close to that in LSA. However in this study, vessels arising from the MCA and the part A1 of ACA were included. Second, some volunteers had history of disorders, which may affect the visualization of LSAs, but the condition was the same for both TOF-MRA and FSBB-MRA. In conclusion, the LSA visualization using TOF-MRA was best with SORS pulse of 400 flip angle than with those of 0 and 750 . However, the optimized TOF-MRA with SORS pulse is not comparable to FSBB-MRA for the visualization of LSA. This indicates that FSBB-MRA would be a better choice at 3T for visualization of the LSA.
ACKNOWLEDGMENTS We express our sincere gratitude to Mrs. Kyoko Takakura, RT and Mr. Hajime Sagawa, RT at Department of Radiology, Kyoto University Hospital and Mr. Naotaka Sakashita at 815
OKUCHI ET AL
Toshiba Medical Systems, Ohtawara-shi, Japan for their help of this study. Grants: This work is funded by a sponsored research program ‘‘Research for Improvement of MR Visualization (No. 150100700014)’’ of Toshiba Medical Systems, Japan, and Grant-in-Aid for Scientific Research on Innovative Areas ‘‘Initiative for High-Dimensional Data-Driven Science through Deepening of Sparse Modeling (No. 4503)’’ of The Ministry of Education, Culture, Sports, Science and Technology, Japan provided to K.T. The former grant contributed to the acquisition of volunteer images of TOF-MRA and FSBB-MRA. The latter grant contributed to analysis of the acquired data. REFERENCES 1. Greenberg SM. Small vessels, big problems. N Engl J Med 2006; 354(14): 1451–1453. 2. Marinkovic SV, Milisavljevic MM, Kovacevic MS, et al. Perforating branches of the middle cerebral artery. Microanatomy and clinical significance of their intracerebral segments. Stroke 1985; 16(6): 1022–1029. 3. Marinkovic S, Gibo H, Milisavljevic M, et al. Anatomic and clinical correlations of the lenticulostriate arteries. Clin Anat 2001; 14(3):190–195. 4. Feekes JA, Hsu SW, Chaloupka JC, et al. Tertiary microvascular territories define lacunar infarcts in the basal ganglia. Ann Neurol 2005; 58(1):18–30. 5. Feekes JA, Cassell MD. The vascular supply of the functional compartments of the human striatum. Brain 2006; 129(Pt 8):2189–2201. 6. Cho ZH, Kang CK, Han JY, et al. Observation of the lenticulostriate arteries in the human brain in vivo using 7.0T MR angiography. Stroke 2008; 39(5): 1604–1606. 7. Kang CK, Park CW, Han JY, et al. Imaging and analysis of lenticulostriate arteries using 7.0-Tesla magnetic resonance angiography. Magn Reson Med 2009; 61(1):136–144. 8. Chen YC, Li MH, Li YH, et al. Analysis of correlation between the number of lenticulostriate arteries and hypertension based on high-resolution MR angiography findings. AJNR Am J Neuroradiol 2011; 32(10):1899–1903. 9. Akashi T, Taoka T, Ochi T, et al. Branching pattern of lenticulostriate arteries observed by MR angiography at 3.0 T. Jpn J Radiol 2012; 30(4): 331–335. 10. Gotoh K, Okada T, Miki Y, et al. Visualization of the lenticulostriate artery with flow-sensitive black-blood acquisition in comparison with time-offlight MR angiography. J Magn Reson Imaging 2009; 29(1):65–69. 11. Okuchi S, Okada T, Ihara M, et al. Visualization of lenticulostriate arteries by flow-sensitive black-blood MR angiography on a 1.5T MRI system: a comparative study between subjects with and without stroke. AJNR Am J Neuroradiol 2013; 34(4):780–784. 12. Miyazaki M, Kojima F, Ichinose N, et al. A novel saturation transfer contrast method for 3D time-of-flight magnetic resonance angiography: a sliceselective off-resonance sinc pulse (SORS) technique. Magn Reson Med 1994; 32(1):52–59.
816
Academic Radiology, Vol 21, No 6, June 2014
13. Admiraal-Behloul1 F, Blink1 E, Zhang1 B, et al. High resolution time-of flight MRA using slice selective saturation transfer contrast and water excitation technique for the visualization of the lenticulostriate arteries at 1.5T. Proc Intl Soc Mag Reson Med 2011; 19:363. 14. Okada T, Miyazaki M, Kido A, Visualization of the lenticulostriate arteries at 3T: a comparative study of TOF-MRA with MTC and FSBB-MRA. Proceedings of the 17th Annual Scientific Meeting of KSMRM. 2012:389. 15. Kundel HL, Polansky M. Measurement of observer agreement. Radiology 2003; 228(2):303–308. 16. Wolff SD, Balaban RS. Magnetization transfer contrast (MTC) and tissue water proton relaxation in vivo. Magn Reson Med 1989; 10(1):135–144. 17. Koenig SH. Cholesterol of myelin is the determinant of gray-white contrast in MRI of brain. Magn Reson Med 1991; 20(2):285–291. 18. Pike GB, Hu BS, Glover GH, et al. Magnetization transfer time-of-flight magnetic resonance angiography. Magn Reson Med 1992; 25(2):372–379. 19. Bradley WG, Jr, Waluch V. Blood flow: magnetic resonance imaging. Radiology 1985; 154(2):443–450. 20. Edelman RR, Mattle HP, Wallner B, et al. Extracranial carotid arteries: evaluation with ‘‘black blood’’ MR angiography. Radiology 1990; 177(1):45–50. 21. Alexander AL, Buswell HR, Sun Y, et al. Intracranial black-blood MR angiography with high-resolution 3D fast spin echo. Magn Reson Med 1998; 40(2):298–310. 22. Mayo JR, Culham JA, MacKay AL, et al. Blood MR signal suppression by preexcitation with inverting pulses. Radiology 1989; 173(1):269–271. 23. Parker DL, Goodrich KC, Masiker M, et al. Improved efficiency in doubleinversion fast spin-echo imaging. Magn Reson Med 2002; 47(5): 1017–1021. 24. Reichenbach JR, Barth M, Haacke EM, et al. High-resolution MR venography at 3.0 Tesla. J Comput Assist Tomogr 2000; 24(6):949–957. 25. Haacke EM, Xu Y, Cheng YC, et al. Susceptibility weighted imaging (SWI). Magn Reson Med 2004; 52(3):612–618. 26. Le Bihan D, Breton E, Lallemand D, et al. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988; 168(2): 497–505. 27. Kimura T, Ikedo M, Takemoto S. Hybrid of opposite-contrast MR angiography (HOP-MRA) combining time-of-flight and flow-sensitive black-blood contrasts. Magn Reson Med 2009; 62(2):450–458. 28. Kang CK, Park CA, Lee H, et al. Hypertension correlates with lenticulostriate arteries visualized by 7T magnetic resonance angiography. Hypertension 2009; 54(5):1050–1056. 29. Kang CK, Park CA, Park CW, et al. Lenticulostriate arteries in chronic stroke patients visualised by 7 T magnetic resonance angiography. Int J Stroke 2010; 5(5):374–380. 30. Gotoh K, Okada T, Satogami N, et al. Evaluation of CT angiography for visualisation of the lenticulostriate artery: difference between normotensive and hypertensive patients. Br J Radiol 2012; 85(1019):e1004–e1008. 31. Sasaki T, Kodama N, Matsumoto M, et al. Blood flow disturbance in perforating arteries attributable to aneurysm surgery. J Neurosurg 2007; 107(1): 60–67. 32. Park DH, Kang SH, Lee JB, et al. Angiographic features, surgical management and outcomes of proximal middle cerebral artery aneurysms. Clin Neurol Neurosurg 2008; 110(6):544–551. 33. Lustig M, Donoho D, Pauly JM. Sparse MRI: the application of compressed sensing for rapid MR imaging. Magn Reson Med 2007; 58(6): 1182–1195.