Surveillance of Unruptured Intracranial Saccular Aneurysms Using Noncontrast 3D-Black-Blood MRI: Comparison of 3D-TOF and Contrast-Enhanced MRA with 3D-DSA

Sixty-four patients with 68 saccular unruptured intracranial aneurysms were recruited. Patients underwent 3T MR imaging with 3D-TOF-MRA, 3D black-blood MR imaging, and contrast-enhanced MRA, and they underwent 3D rotational angiography within 2 weeks. The neck, width, and height of the unruptured intracranial aneurysms were measured by 2 radiologists independently on 3D rotational angiography and 3 MR imaging sequences. 3D black-blood MR imaging demonstrates the best agreement with DSA, with the smallest limits of agreement and measurement error. 3D-TOF-MRA had the largest limits of agreement and measurement error. The authors conclude that 3D black-blood MR imaging achieves better accuracy for aneurysm size measurements compared with 3D-TOF, using 3D rotational angiography as a criterion standard. BACKGROUND AND PURPOSE: Patients with unruptured intracranial aneurysms routinely undergo surveillance imaging to monitor growth. Angiography is the criterion standard for aneurysm diagnosis, but it is invasive. This study aimed to evaluate the accuracy and reproducibility of a 3D noncontrast black-blood MR imaging technique for unruptured intracranial aneurysm measurement in comparison with 3D-TOF and contrast-enhanced MRA, using 3D rotational angiography as a reference standard. MATERIALS AND METHODS: Sixty-four patients (57.3 ± 10.9 years of age, 41 women) with 68 saccular unruptured intracranial aneurysms were recruited. Patients underwent 3T MR imaging with 3D-TOF-MRA, 3D black-blood MR imaging, and contrast-enhanced MRA, and they underwent 3D rotational angiography within 2 weeks. The neck, width, and height of the unruptured intracranial aneurysms were measured by 2 radiologists independently on 3D rotational angiography and 3 MR imaging sequences. The accuracy and reproducibility were evaluated by Bland-Altman plots, the coefficient of variance, and the intraclass correlation coefficient. RESULTS: 3D black-blood MR imaging demonstrates the best agreement with DSA, with the smallest limits of agreement and measurement error (coefficients of variance range, 5.87%–7.04%). 3D-TOF-MRA had the largest limits of agreement and measurement error (coefficients of variance range, 12.73%–15.78%). The average coefficient of variance was 6.26% for 3D black-blood MR imaging, 7.03% for contrast-enhanced MRA, and 15.54% for TOF-MRA. No bias was found among 3 MR imaging sequences compared with 3D rotational angiography. All 3 MR imaging sequences had excellent interreader agreement (intraclass correlation coefficient, >0.95). 3D black-blood MR imaging performed the best for patients with intraluminal thrombus (n = 10). CONCLUSIONS: 3D black-blood MR imaging achieves better accuracy for aneurysm size measurements compared with 3D-TOF, using 3D rotational angiography as a criterion standard. This noncontrast technique is promising for surveillance of unruptured intracranial aneurysms.

ABBREVIATIONS: BB ϭ black-blood; CE ϭ contrast-enhanced; CV ϭ coefficient of variance; 3DRA ϭ 3D rotational angiography; IA ϭ intracranial aneurysm; ICC ϭ intraclass correlation coefficient; LOA ϭ limit of agreement; SPACE ϭ spatial and chemical-shift encoded excitation; UIA ϭ unruptured intracranial aneurysm I ntracranial aneurysms (IAs) are common, with approximately 3% of adults having an unruptured IA (UIA). 1 UIAs are associ-ated with morbidity and mortality due to the risk of rupture and the resultant subarachnoid hemorrhage. 2 Of patients surviving ruptured IAs, 30% will experience high morbidity due to associated intraparenchymal hemorrhage from late-onset complications such as vasospasm leading to cerebral infarction. Only approximately one-third of patients with subarachnoid hemorrhage related to IA rupture will fully return to their prior jobs, 3,4 but diagnosis of IAs before rupture may improve patient outcomes. 5 Clinical management of UIAs is based largely on aneurysm size and other features such as location, shape, and sac-to-neck ratio, with saccular aneurysms being the most common type. [6][7][8][9] Because intervention-associated risks may exceed the low rupture rates for smaller UIAs, 5 most of those small aneurysms are followed with surveillance instead of immediate treatment. In addition, more incidental UIAs have been found on widely used noninvasive imaging techniques. Usually, incidental UIAs that are regular in shape and Ͻ7 mm are followed up to assess changes with time that may modify clinical decision-making, 10 but how to best diagnose and monitor UIAs remains uncertain.
Although 3D rotational angiography (3DRA) is the criterion standard for diagnosing IAs as well as evaluating IA morphology measurements, 11 both MR imaging and CTA have become increasingly important in evaluating UIAs. This change is largely because both MR imaging and CTA are less invasive than 3DRA, reducing the risks of IA screening and monitoring of known UIAs. 12-14 However, unlike MR imaging, both 3DRA and CTA expose patients to ionizing radiation and iodinated contrast media. MR imaging is an option that removes the uncertain risks of repeat radiation and contrast media exposure.
Vascular MR imaging has substantially improved with the development of techniques enabling the diagnosis, visualization, and assessment of intracranial aneurysms with high accuracy. 15 Numerous studies have assessed the value of both flow-based (TOF) and contrast-enhanced (CE) MRA techniques. 5 Fewer studies have evaluated 3D noncontrast black-blood (BB) MR imaging in aneurysm characterization, though early data have shown promise 15 with the use of high-isotropic resolution (up to 0.5 mm) to visualize the aneurysm geometry and wall. 16 3D-BB MR imaging, however, has not been validated against 3DRA for aneurysm size measurements or compared with other clinical MRA techniques. This study aimed to compare black-blood MR imaging with 3D-TOF/CE-MRA for aneurysm size measurements using 3DRA as a reference standard. We hypothesized that 3D noncontrast BB MR imaging can be used to measure aneurysm size accurately.

Study Population
This study was approved by the institutional review board of Changhai Hospital. Informed consent was obtained from all patients. A total of 69 consecutive patients with confirmed saccular IAs were recruited between January 2016 and March 2018. All patients underwent 3T MR imaging and 3DRA within 2 weeks.
3DRA. 3DRA was performed on the Artis zee Biplane angiographic system (VC14; Siemens). A 5-second DSA acquisition protocol was performed with contrast injection at a rate of 3 mL/s of Ultravist (iopromide; Bayer HealthCare, Berlin, Germany). During the 5-second acquisition after a delay of 1 second, a 200°r otation of the C-arm was performed to obtain 133 frames. The parameters were as follows: FOV ϭ 320 ϫ 320 mm 2 , matrix ϭ 1024 ϫ 1024.

Image Analysis
Image analysis was performed independently by 2 radiologists (X.W. and B.T. with 5 and 9 years of experience) who were blinded to patients' clinical data. All measurements were conducted on the postprocessing workstation (syngo X Workspace; Siemens). Measurement of 3DRA was performed on volume-rendered reconstruction, and measurements of TOF-MRA and CE-MRA were performed on MIP reconstructions. The MIP images were imported to the workstation, and the image viewer automatically set a default window/level for the best display. In addition, 2 observers were also allowed to adjust the window/level and projection angles to best display the aneurysm and the parent vessel. Measurement of 3D-BB MR imaging was performed on MPR images.
For each IA, we measured the following parameters: 1) height: the maximum distance from the neck center to the dome of the aneurysm; 2) width: the longest diameter of an aneurysm perpendicular to the height; and 3) neck: the minimum width of the aneurysm at its junction with the parent vessel (Fig 1). Measurements were performed by the 2 readers (X.W. and B.T.) separately, with an interval of 2 weeks between different modalities. One reader performed the second measurement (X.W.) 2 weeks later using the same method to evaluate the intrareader variability. All these measurements were performed only for the luminal geometry, for which DSA could be used as a reference standard. If the aneurysms had intraluminal thrombus, the images were evaluated separately for the true aneurysm geometry.

Statistical Analysis
Normality assumptions were assessed using the Shapiro-Wilk test. Continuous variables were expressed as either mean Ϯ SD or median (interquartile range). Categoric variables were summarized by count (percentage). Comparisons of aneurysm height, width, and neck diameter among different imaging modalities were analyzed using the Kruskal-Wallis test. Measurement differences between MR imaging modalities and DSA as well as intraand interreader variabilities were assessed using the Bland-Altman analysis. Bias was assessed as the mean of the pair differences, and the 95% limits of agreement (LOA) were defined as bias Ϯ 1.96 ϫ SD. Measurement error was quantified by the coefficient of variance (CV ϭ SD / mean ϫ 100%). Agreement between measurements was summarized by the intraclass correlation coefficient (ICC) with a 2-way mixed model. An ICC value Ͼ0.80 indicated excellent agreement. On the basis of the measurement errors of each MR imaging sequence, the sample sizes needed to detect 5%, 10%, and 20% changes in aneurysm size growth in longitudinal studies were calculated by using 0.9 power and .05 significance as described in a previous publication. 17 Statistical analyses were performed using SPSS, Version 24.0 (IBM, Ar-monk, New York) and MedCalc for Windows, Version 9.4.2.0 (MedCalc Software, Mariakerke, Belgium).

RESULTS
Of the 69 recruited patients, 5 were excluded for the following reasons: 1) an interval of Ͼ2 weeks between MR imaging and DSA (n ϭ 2); 2) a ruptured aneurysm (n ϭ 1); and 3) insufficient image quality due to motion artifacts on 3D-BB MR imaging (n ϭ 2). The remaining 64 patients (57.3 Ϯ 10.9 years of age, 41 women) with 68 saccular UIAs were included. UIAs were located in the internal carotid artery (n ϭ 32), middle cerebral artery (n ϭ 12), anterior communicating artery (n ϭ 10), posterior communicating artery and posterior cerebral artery (n ϭ 8), and vertebralbasilar artery (n ϭ 6). The average interval between MR imaging and DSA was 2.2 Ϯ 1.6 days (range, 1-9 days). Mild flow artifacts were observed in 3 UIAs with aneurysm diameters of Ͼ15 mm on 3D-BB MR imaging. These artifacts did not affect the measurement accuracy.

Agreement between MR Imaging Sequences and DSA
As shown in Table 1, there was no significant difference in aneurysm height, width, and neck diameter among TOF/CE-MRA, 3D-BB MR imaging, and DSA (P Ͼ .05). Agreement of the quantitative measurements between MR imaging modalities and DSA is summarized in Table 2. Bland-Altman plots for each measurement are shown in Fig 2 and the On-line Figure. Measurements between MR imaging modalities and DSA showed excellent agreement with ICCs of Ͼ0.96. Measurement on TOF-MRA showed the largest variance (overall CV, 15.54%) and higher LOA compared with measurements on CE-MRA (overall CV, 7.03%) and 3D-BB MR imaging (overall CV, 6.26%). Based on the measurement errors of each MR image, the calculated sample sizes in future longitudinal studies to detect 5%, 10%, and 20% change in aneurysm growth are shown in Table 3. Compared with 3D-TOF, the use of CE-MRA or 3D-BB MR imaging reduces the sample size    Figure 3 shows that TOF is limited as a means of displaying the aneurysm geometry due to severe flow artifacts, while DSA, CE-MRA, and SPACE characterize the geometry nicely. Figure 4 shows an aneurysm with intraluminal thrombus. 3D-BB MR imaging can clearly show the entire aneurysm structure including both the lumen and the thrombus.

Intra-and Interreader Agreement
Intra-and interreader agreement is shown in Table 4. There were no significant mean differences between readers (P Ͼ .05). The overall intra-and interreader measurements showed excellent agreement, with ICCs of Ͼ0.97.

Aneurysms with Intraluminal Thrombus
Ten aneurysms were found with intraluminal thrombus: 3 aneurysms located in the middle cerebral artery, 2 located in the internal carotid artery, 2 located in the anterior communicating artery, 2 located in the posterior communicating artery, and 1 located in the basilar artery. Aneurysm sizes ranged from 8.0 to 28.2 mm. 3D-BB MR imaging gave the best evaluation of aneurysm dimensions, with clear delimitation of both the lumen and thrombus (Fig 3). The median thickness of the intraluminal thrombus was 2.8 mm (IQR, 0.8 -4.4 mm).

DISCUSSION
In this study, measurements obtained on 3D-BB MR imaging were in close agreement with 3DRA measurements and had a better accuracy than 3D-TOF with less than half the TOF measurement error. The sample size needed to measure aneurysm size change in longitudinal studies would be reduced by 84% using 3D-BB MR imaging compared with using 3D-TOF. Our results support the use of noncontrast 3D-BB MR imaging for intracranial aneurysm surveillance. Despite previous studies evaluating the use of BB MR imaging sequences in the characterization of IAs, this is the first study to validate 3D-BB MR imaging against the criterion standard 3DRA for aneurysm morphology measurements. In addition, this is the first study to compare 3D-BB MR imaging with 3D-TOF and CE-MRA for IA measurement accuracy. Although 3DRA remains the criterion standard in aneurysm evaluation and treatment planning with high spatial resolution and excellent hemodynamic temporal resolution, 11 DSA is inva-sive with procedural risks such as arterial dissection, hematomas, stroke, and even death that are more important than the small theoretic risks of radiation and contrast exposure. 18 Alternatively, CTA is a fast, noninvasive imaging test that is often used in first-line aneurysm assessment. CTA has widespread availability and has been shown to be of benefit in not only aneurysm detection [19][20][21][22][23] but also presurgical planning. 24 Despite the advantages of CTA, like DSA, CTA uses radiation and iodinated contrast, reducing its adoption in UIA screening and monitoring. 12 MR imaging does not have these issues. 25 Other disadvantages of CTA include the fact that despite the development of improved techniques in the region of the skull base, 26,27 MR imaging lacks the attenuation and streak artifacts occasionally encountered in CTA. 28 MR imaging can play a major role in UIA detection, monitoring, and treatment planning. 29 3D-TOF and CE-MRA are commonly used methods for UIA evaluation. 3D-TOF-MRA is a popular method because it does not require a contrast agent 10,30,31 ; however, it has flow artifacts and can confuse T1-weighted hyperintense thrombus within an aneurysm with flow-related enhancement. 32,33 Bright thrombus can mimic flowing lumen, and isointense thrombus is difficult to distinguish from surrounding brain tissue. 15 TOF-MRA can also make detection of IAs difficult in the setting of T1 hyperintense intraparenchymal or subarachnoid hemorrhage. [34][35][36] Our study observed that while still similar to 3DRA, 3D-TOF-MRA had the least accuracy of the MR imaging-based methods tested, in accordance with previous literature. CE-MRA is an accurate method for UIA evaluation because it is not flow-dependent. 37 CE-MRA was shown to be superior to 3D-TOF-MRA for assessment of sac shape, visualizing an aneurysm neck, calculating of the sac/neck ratio, and identifying branches originating from the sac and/or neck, 5 especially for large aneurysms (Ͼ13 mm), when flow artifacts were commonly present. Despite these advantages of CE-MRA, its limitations include the use of gadolinium-based contrast media, dependence on timing of contrast administration that may lead to venous contamination, and radiofrequency power deposition at high field strengths. 38 Recent concerns regarding gadolinium deposition in the brain parenchyma, while uncertain and controversial, call into question its widespread use in repeat imaging for surveillance. 39 When we compared it with other imaging modalities, 3D-BB MR imaging has clear advantages. First, it is an accurate method for UIA measurements as evidenced in the superior agreement of this study with DSA, which was comparable with CE-MRA. Although our CE-MRA protocol had better resolution (around 0.7 mm isotropic) than most routine clinical protocols (0.9 -1.1 mm), 5 3D-BB MR imaging had an even higher resolution (0.5 mm isotropic), which may explain its slightly improved accuracy. 3D-BB MR imaging has also been used for intracranial stenosis quantification 11 and extracranial carotid plaque imaging 40 in studies showing excellent agreement with DSA. Our study extends these observations in showing that 3D-BB MR imaging provides accurate assessment of aneurysms. The high accuracy of 3D-BB MR imaging is favorable for future longitudinal studies, which can reduce sample size more significantly than with 3D-TOF and reduce the cost of the clinical trial. Accurate quantification of aneurysm growth is important because fast-growing aneurysms have a much higher risk of rupture. 41,42 Second, as a noncontrast acquisition, 3D-BB MR imaging is ideal for repeat, routine surveillance of UIAs compared with contrast techniques including CE-MRA and CTA. It can also be repeated if the first acquisition fails due to patient motion or other issues. Third, 3D-BB SPACE is insensitive to flow artifacts, which is superior to 3D-TOF (Figs 3 and 4). The 3D fast-spin-echo BB sequence is inherently blood-suppressed. 43 Although in large aneurysms, flow artifacts can still occur, several additional techniques, including improved motion-sensitized driven-equilibrium 44 and delay alternating with nutation for tailored excitation (DANTE) 45 may be used to optimize blood suppression. Previous work comparing 3D-TOF and BB techniques showed significant advantages to BB imaging in cases of large aneurysms or slow blood flow. 36,46 In 1 study, the authors found 3 patients in whom the aneurysm was not demonstrated on 3D-TOF due to the presence of local hematoma, but by performing a BB sequence, they found that all of the IAs were subsequently identified without the need for contrast. 15 Last, 3D-BB MR imaging can also visualize intraluminal thrombus and the vessel wall, which is a unique advantage in comparison with angiographic methods. Because vessel wall features and postcontrast enhancement have been studied as potential markers of aneurysm rupture, 25,47 this ability could become increasingly important in aneurysm characterization and monitoring.

Limitations
There are some limitations to this study. First, we evaluated only saccular aneurysms with fusiform aneurysms excluded. Second, our current 3D-BB MR imaging sequence has a scan time of approximately 9 minutes, greater than the time it takes for either CE-MRA (2-3 minutes) or TOF (ϳ5 minutes). It is possible that a lower resolution (0.6 -0.7 mm isotropic) can achieve sufficient accuracy with a significantly reduced scan time. Advanced acceleration techniques, including compressed sensing, may potentially reduce the scan time significantly with similar image quality. 48 Future studies are needed to determine the minimum spatial resolution necessary and apply new acceleration techniques clinically. Third, the image-reconstruction methods were not the same across imaging modalities. We used the reconstruction methods that were conventionally used in clinical routine, 15,49,50 and also note that the influence of reconstruction methods for aneurysm measurements has been rarely studied. However, we believe the differences caused by reconstruction methods were much less than the differences caused by imaging techniques, as shown in our data.
Despite these limitations, we believe that they are offset by the improved IA measurements compared with the other MR imaging techniques as well as the ability of 3D-BB MR imaging to characterize the vessel wall without the use of gadolinium-based contrast, supporting the use of noncontrast BB MR imaging to replace CE-MRA in monitoring IAs.

CONCLUSIONS
3D-BB MR imaging achieves superior accuracy for intracranial aneurysm size measurements over 3D-TOF using 3DRA as the criterion standard, though all MR imaging measurements did not significantly differ from those with 3DRA. This noncontrast technique is promising for clinical surveillance of patients with unruptured intracranial aneurysms.