Professional Documents
Culture Documents
Jurnal 4
Jurnal 4
org
q
RSNA, 2017
q
RSNA, 2017
A
cute subarachnoid hemorrhage features of cerebral aneurysms and the symptom deterioration that occurred
caused by rupture of a cerebral associated circulation is critical for the while the patient was waiting or moving
aneurysm is associated with high selection and optimization of the en- locations for MR angiography and that
morbidity and mortality, and effective dovascular or surgical technique. Few the neurologist believed was caused by
treatment depends on efficient and studies have focused on the capability rebleeding (n = 12).
accurate detection of the ruptured an- of 3D TOF MR angiography, as com- The patients directly underwent
eurysm (1,2). Conventional catheter pared with conventional angiography, digital subtraction angiographic (DSA)
angiography is generally considered for detailed morphologic display of ce- embolism (n = 9) or surgical clipping
the reference-standard technique for rebral aneurysms. (n = 3) after emergent consultation
detection of cerebral aneurysms, but In this study, we aimed to evaluate with the neurointerventional radiolo-
noninvasive methods, such as com- the diagnostic accuracy of aneurysm gist and neurosurgeon. Among the re-
puted tomographic (CT) angiography or detection and the fidelity of morpho- maining 287 patients, satisfactory MR
magnetic resonance (MR) angiography, logic characterization of 3D TOF MR angiographic images were successfully
have become promising alternatives angiography at 3.0 T in patients with acquired in 98.3% (282 of 287). Motion
following the development of improved a Glasgow coma score of 15 and non- artifacts, leading to uninterpretable im-
equipment and imaging sequences (3– contrast material–enhanced CT findings age quality, occurred in 15 patients.
5). CT angiography is recommended of acute nontraumatic subarachnoid In 10 of these 15 patients, a second
as the first choice because of its fast hemorrhage. examination was successful, with im-
scanning speed and convenience; how- age quality sufficient for diagnosis.
ever, it is associated with exposure to However, the other five patients were
ionizing radiation and nephrotoxicity Materials and Methods also intolerant of the procedure the
risks. MR angiography is free of these second time and were finally excluded
problems and is therefore a promising Patients from the study. Meanwhile, another
alternative. Our previous studies have The institutional review board approved five patients declined to undergo DSA
revealed that the diagnostic accuracy the study protocol, and patients or their and were also excluded. No patient ex-
of MR angiography for the detection relatives gave informed consent before perienced rebleeding during MR angi-
of cerebral aneurysm can be as great their participation in this prospective ography. Thus, as the study flowchart
as 96%–97% with the use of a three- study. This discussion about the study (Fig 1) shows, a total of 277 patients
dimensional (3D) time-of-flight (TOF) was held immediately after the unen- who underwent both 3D TOF MR an-
MR angiographic technique (6,7). Ac- hanced CT examination was finished giography and DSA were included in
curate evaluation of the morphologic and involved undergoing MR angiogra- the final analysis. The study group com-
phy instead of CT angiography. prised 117 men and 160 women, with
Advances in Knowledge Between February 1, 2009, and Au- a mean age of 53.87 years 6 11.87
nn Three-dimensional time-of-flight gust 31, 2015, a total of 545 patients (standard deviation) (age range, 24–79
(TOF) MR angiography at 3.0 T had nontraumatic subarachnoid hemor-
has high accuracy and sensi- rhage that was confirmed with nonen-
https://doi.org/10.1148/radiol.2017161469
tivity—96.8% (268 of 277) and hanced CT in our hospital. Of the 391
98.2% (219 of 223), respec- patients with subarachnoid hemorrhage Content codes:
tively—for the detection of and a Glasgow coma score of 15, 314
Radiology 2017; 284:191–199
cerebral aneurysms in patients were eligible and agreed to inclusion in
with subarachnoid hemorrhage this study. Twenty-seven patients were Abbreviations:
excluded for the following reasons: (a) DSA = digital subtraction angiography
who have a Glasgow coma score ICA = internal carotid artery
of 15. pacemaker or steel implants (n = 10),
3D = three-dimensional
(b) allergy to contrast material (n = 1),
nn Three-dimensional TOF MR an- TOF = time of flight
(c) renal dysfunction that precluded the
giographic images were usable use of contrast material (n = 4), and (d) Author contributions:
for reconstruction in 98.3% of Guarantor of integrity of entire study, Y.Z.; study concepts/
patients (282 of 287). study design or data acquisition or data analysis/inter-
nn Accurate display of morphologic Implication for Patient Care pretation, all authors; manuscript drafting or manuscript
revision for important intellectual content, all authors; man-
features at three-dimensional nn Three-dimensional TOF MR angi- uscript final version approval of final version of submitted
TOF MR angiography was 92.5% ography can efficiently depict manuscript, all authors; agrees to ensure any questions
(236 of 255) for size, 86.3% (220 cerebral aneurysms and accu- related to the work are appropriately resolved, all authors;
of 255) for neck width, 94.5% rately display the morphologic literature research, M.L., Y.Z., Y.C.; clinical studies, M.L.,
(241 of 255) for shape, and structure in stable patients with Y.Z., H.S., B.G., H.L., Y.L., H.T.; statistical analysis, M.L., Y.Z.;
and manuscript editing, M.L., Y.Z., Y.C.
96.9% (247 of 255) for relation- subarachnoid hemorrhage and a
ship to adjacent vessel. Glasgow coma score of 15. Conflicts of interest are listed at the end of this article.
Jiangsu, China) was injected for a total Aneurysm size and neck width were categoric variables. Patient-by-patient
of 10 mL for the ICA (rate, 4–5 mL/ recorded as the maximum two-dimen- and aneurysm-by-aneurysm analyses
sec), 7 mL for the vertebral artery sional angiographic dimension. Errors of were performed. The diagnostic perfor-
(rate, 2–3 mL/sec), and 10–15 mL per less than 15% for maximum aneurysm mance of 3D TOF MR angiography was
artery for rotational angiography (rate, size and 10% for aneurysm neck width summarized in terms of the overall accu-
2–3 mL/sec). Two observers (reviewers was considered acceptable, considering racy, sensitivity, specificity, positive pre-
H.T. and H.L., with 17 and 10 years of that measurement was made by using dictive value, negative predictive value,
experience, respectively, in interven- data from MR angiographic and DSA and area under the receiver operating
tional neuroradiology) evaluated the sources and by different persons. characteristic curve. For patient-based
cerebral aneurysms together. An aneurysm was considered giant evaluation, if a patient had multiple an-
if the greatest diameter was 25 mm or eurysms and one of the aneurysms was
Image Review larger, large if it was 10–25 mm, small if missed or overdiagnosed, this patient
Three observers (M.L. [observer A], it was 3.0–10 mm, and tiny if it was less was considered to have false-negative
Y.L. [observer B], and Y.Z. [observer than 3.0 mm. The aneurysm neck was or false-positive findings. The 95% con-
C], with 20, 9, and 8 years of experi- considered wide if it was larger than fidence interval of each index was cal-
ence, respectively, in interventional 4 mm or if the fundus-to-neck ratio culated by using the score confidence
neuroradiology) who were blinded to was smaller than 2. For assessment of interval (10). Multivariate logistic regres-
clinical findings and DSA results inde- morphologic features of the aneurysm, sion analysis was performed to identify
pendently analyzed the 3D TOF MR three features were considered: the the variables that significantly affected
angiography with volume-rendering im- presence of a daughter sac, the pres- the morphologic assessment with 3D
age data sets at an offline workstation ence of lobulation, and the margin (9). TOF MR angiography. All analyses were
from multiple on-screen viewing angles A daughter sac was deemed to be pre- performed by using statistical software
by using the single-artery highlighting sent if two-dimensional or 3D MR an- (SPSS 16.0; SPSS, Chicago, Ill), and P
approach. The source images and max- giographic and/or DSA images revealed , .05 was considered to indicate a sig-
imum intensity projection images were an irregular protrusion from the wall of nificant difference.
presented on screen, allowing for ad- the aneurysm that was less than 25% of
justment of the appropriate threshold the total volume of the main sac. Lob-
of the window width and window level ulation was defined as a protuberance Results
to diagnose or differentiate aneurysms arising directly from the primary neck
with infundibula. Cases with one or of the aneurysm or from the main body Aneurysm Detection with 3D TOF MR
more aneurysms detected were consid- and representing 25% or more of the Angiography
ered positive; all others were consid- volume of the main sac. The margin of DSA images revealed 265 aneurysms
ered negative (regardless of other ce- the aneurysm was classified as a smooth in 225 patients (87 men [mean age,
rebral vascular disease). For analysis of or irregular shape. With regard to the 57.08 years 6 13.19; range, 25–84
the morphologic assessment of 3D TOF relationship between aneurysm and years]; 138 women [mean age, 54.21
MR angiography, the average of the parent artery, there were two grades: years 6 12.25; range, 27–79 years];
values of the three reviewers was used (a) aneurysm arising from the side wall P = .10). Altogether, five patients had
for aneurysm size and neck width. An- of the artery and (b) aneurysm arising three aneurysms each, 30 patients had
eurysm configuration and relationship from an arterial bifurcation. The latter two aneurysms each, and 190 patients
to adjacent arteries were decided by could be of three types: (a) aneurysm had one aneurysm each. In patient-
consensus among the three reviewers. neck arising from the angle of the bifur- based evaluations, 3D TOF MR an-
cation, (b) more than 75% of the aneu- giography yielded accuracy of 96.8%
Definitions rysm neck arising from one limb of the (268 of 277), sensitivity of 98.2% (219
We considered an aneurysm to be a bifurcation, and (c) aneurysm involving of 223), specificity of 91% (49 of 54),
saccular protrusion from the side wall the entire angle of the bifurcation. Pre- positive predictive value of 97.8% (219
or from a bifurcation of the cerebral cise morphologic display of the cerebral of 224), and negative predictive value
arteries, without the artery emerging at aneurysm at 3D TOF MR angiography of 92% (49 of 53) (mean k = 0.959;
its top. Aneurysm locations were clas- was considered a failure if any of the range, 0.951–0.964). In aneurysm-
sified as the ICA, including the poste- morphologic parameters—location, based evaluations, 3D TOF MR an-
rior communicating artery; the anterior size, neck width, shape, and relation giography yielded accuracy of 96.6%
cerebral artery, including the anterior to adjacent artery—did not conform to (309 of 320), sensitivity of 98.1% (260
communicating artery; the middle cere- the DSA findings. of 265), specificity of 89% (49 of 55),
bral artery, including the M1–2 segment; positive predictive value of 97.7% (260
and the vertebral and basilar arteries, Statistical Analysis of 266), and negative predictive value
including the vertebral, basilar, cere- Interobserver agreement was expressed of 91% (49 of 54) (mean k = 0.947;
bellar, and posterior cerebral arteries. in terms of the Cohen k value for range, 0.932–0.967). On the 3D TOF
Table 1
Diagnostic Performance of 3D TOF MR Angiography in Patient- and Aneurysm-based Evaluations
Evaluation and Observer Sensitivity (%)* Specificity (%) PPV (%) NPV (%) Accuracy (%)
Patient-based evaluations†
Observer A 99.6 [0.976, 0.999] 90.4 [0.794, 0.958] 97.8 [0.950, 0.991] 97.9 [0.891, 0.996] 97.8 [0.953, 0.990]
(224/225) (47/52) (224/229) (47/48) (271/277)
Observer B 98.7 [0.962, 0.996] 94.2 [0.843, 0.980] 98.7 [0.962, 0.996] 94.2 [0.843, 0.980] 97.8 [0.953, 0.990]
(222/225) (49/52) (222/225) (49/52) (271/277)
Observer C 99.1 [0.968, 0.998] 94.2 [0.843, 0.980] 98.7 [0.962, 0.996] 96.1 [0.868, 0.989] 98.2 [0.959, 0.992]
(223/225) (49/52) (223/226) (49/51) (272/277)
Aneurysm-based evaluations‡
Observer A 99.6 [0.979, 0.999] 90.9 [0.804, 0.960] 98.1 [0.957, 0.992] 98.0 [0.896, 0.996] 98.1 [0.959, 0.991]
(264/265) (50/55) (264/269) (50/51) (314/320)
Observer B 98.9 [0.968, 0.996] 92.7 [0.827, 0.971] 98.5 [0.962, 0.994] 94.4 [0.848, 0.981] 97.8 [0.955, 0.989]
(262/265) (51/55) (262/266) (51/54) (313/320)
Observer C 98.9 [0.968, 0.996] 94.5 [0.851, 0.981] 98.8 [0.968, 0.996] 94.5 [0.851, 0.981] 98.1 [0.959, 0.991]
(262/265) (52/55) (262/265) (52/55) (314/320)
Note.—Unless otherwise indicated, data in parentheses are raw data. AUC = area under the receiver operating characteristic curve, NPV = negative predictive value, PPV = positive predictive value.
* Data in brackets are 95% confidence intervals.
†
k Values (coefficients of interreader agreement) ranged from 0.951 to 0.964.
‡
k Values (coefficients of interreader agreement) ranged from 0.932 to 0.967.
Figure 2
Figure 2: A ruptured aneurysm in the C6 segment of the left ICA in a 46-year-old woman. A, Three-dimensional TOF MR angiogram shows a small aneurysm
(arrow) with a daughter sac (arrowhead) at the C6 segment of the left ICA. B, DSA image with volume-rendering reconstruction and, C, two-dimensional DSA image
show the aneurysm (arrow) and its daughter sac (arrowhead) seen at 3D TOF MR angiography with volume-rendering reconstruction. D, DSA image shows total
occlusion of the aneurysm (arrow) after embolization with coils.
Figure 3
Figure 3: A ruptured aneurysm in the anterior communicating artery in a 41-year-old man. A, Three-dimensional TOF MR angiogram shows a small aneurysm
(arrow) with a lobulated shape arising from the anterior communicating artery. B, DSA image with volume-rendering reconstruction shows the aneurysm (arrow) seen
at 3D TOF MR angiography with volume-rendering reconstruction. C, Two-dimensional DSA image shows an aneurysm (arrow) in the anterior communicating artery.
D, DSA image shows total occlusion of the aneurysm (arrow) after embolization with coils.
within 10% of the neck width as mea- confidence interval: 1.062, 3.955; P 0.731; P = .004) was the only variable
sured at DSA in 86.3% of aneurysms = .032) and neck width (odds ratio, associated with the accuracy of mor-
(220 of 255), and accurately character- 0.378; 95% confidence interval: 0.195, phologic assessment of 3D TOF MR
ized shape in 94.5% aneurysms (241 0.731; P = .004) measurements at 3D angiography.
of 255) and relation to adjacent vessel TOF MR angiography were significantly
in 96.9% aneurysms (247 of 255). For different from the DSA values. This
the 35 aneurysms with incorrectly dis- analysis indicated that the larger size Discussion
played neck size, 94% (33 of 35) were of the aneurysm body or smaller aneu- The major findings of our study revealed
overestimated. Mean aneurysm neck rysm neck width was associated with that 3D TOF MR angiography at 3.0 T
width was 3.17 mm 6 2.28 (range, a greater chance of inaccurate display has a sensitivity of more than 95% for
0.6–33.7 mm) displayed at MR angiog- of the aneurysm morphologic features detection of aneurysms in patients with
raphy versus 2.96 mm 6 2.32 (range, at 3D TOF MR angiography. Multiple subarachnoid hemorrhage who have a
0.6–35.0 mm) at DSA (P , .01) (Fig 5). logistic regression analysis revealed Glasgow coma score of 15. In our study
Univariate analysis revealed that an- that aneurysm neck width (odds ratio, population, the procedure was success-
eurysm size (odds ratio, 2.050; 95% 0.378; 95% confidence interval: 0.195, fully performed in 98.3% of patients.
Figure 4
Figure 4: A ruptured aneurysm in the M1 segment of the right middle cerebral artery in a 42-year-old man. A, Three-dimensional
TOF MR angiogram shows a small aneurysm (arrow) with an irregular margin at the bifurcation site of the M1 segment of the right
middle cerebral artery. B, DSA image with volume-rendering reconstruction shows the aneurysm (arrow) seen at 3D TOF MR angiogra-
phy with volume-rendering reconstruction. C, Two-dimensional DSA image shows an aneurysm (arrow) in the M1 segment of the right
middle cerebral artery.
Previous studies grouped ruptured and awake patients (Glasgow coma score of must also be considered in these pa-
unruptured aneurysms together when 15) who could fully cooperate during tients; during our study, the condition
investigating the diagnostic perfor- the examination. Despite this, imaging of 12 patients decompensated during
mance of MR angiography. In our study, attempts for the second examination this time period, and they proceeded
we focused on ruptured aneurysms to failed in five patients; these patients to surgery instead of imaging, and an
clarify the following issues: (a) whether were excluded from the study because additional five patients were unable to
the diagnostic performance of MR an- of safety concerns. complete the MR imaging examination.
giography is similar to that of DSA, (b) Previous CT angiographic or MR In conclusion, 3D TOF MR angiog-
whether MR angiography is feasible angiographic studies were usually per- raphy has high diagnostic accuracy in
and safe in patients with subarachnoid formed only for detection of aneu- the detection of ruptured cerebral an-
hemorrhage, and (c) whether MR angi- rysms. However, with superior spatial eurysms in selected patients with sub-
ography can provide accurate prepro- resolution and better visualization of arachnoid hemorrhage (stable Glasgow
cedure geometric information to guide distal vessels by using 3.0-T MR imag- coma score of 15), and its accuracy
surgical or endovascular treatment. ing, detailed and reliable information in displaying the morphologic features
The performance of 3D TOF MR an- on the morphologic features of an an- of aneurysms appears to be similar to
giography in the detection of ruptured eurysm can be obtained, which can be that of DSA. We recommend the use
aneurysms is likely to be inferior to its useful for treatment planning (15,16). of 3D TOF MR angiography as a tool
performance in cases of unruptured In the case of a ruptured aneurysm, the for screening for ruptured cerebral an-
aneurysms because hemorrhage in the display of a daughter sac or lobulation eurysm and for treatment planning be-
subarachnoid space may adversely af- can help locate the hemorrhage site, cause of its noninvasive nature and its
fect aneurysm display (6,7). However, accurate display of the aneurysm neck high accuracy and sensitivity.
we found that 3D TOF MR angiography can help plan endovascular remodeling
Disclosures of Conflicts of Interest: M.L. dis-
had an accuracy of 97.9%, sensitivity of or clipping techniques, and accurate closed no relevant relationships. Y.Z. disclosed
99.1%, and specificity of 92.9%, which display of the relationship between the no relevant relationships. H.S. disclosed no rele-
is similar to the diagnostic performance aneurysm sac and the adjacent artery vant relationships. B.G. disclosed no relevant re-
lationships. H.L. disclosed no relevant relation-
of CT angiography for the detection of can help to evaluate ischemia risk dur- ships. Y.L. disclosed no relevant relationships.
ruptured aneurysm (4). More interest- ing coiling or clipping. Aneurysm neck H.T. disclosed no relevant relationships. Y.C.
ingly, we found that with 3D TOF MR was found to be the only parameter that disclosed no relevant relationships.
angiography, overdiagnosis is more could not be accurately assessed with
likely than missed diagnosis. This may 3D TOF MR angiography according to
have been because (a) high spatial and multivariate logistic regression analysis. References
temporal resolution was achieved with This is probably because measurement 1. Molyneux A, Kerr R, Stratton I, et al. In-
the application of high-magnetic-field of data from different sources will be ternational Subarachnoid Aneurysm Trial
MR imaging and the previously opti- associated with some degree of error. It (ISAT) of neurosurgical clipping versus en-
mized imaging parameters and (b) 3D is also possible that when the aneurysm dovascular coiling in 2143 patients with rup-
tured intracranial aneurysms: a randomised
volume-rendering reconstruction and sac tightly touches the parent artery, a
trial. Lancet 2002;360(9342):1267–1274.
the single-artery-highlighting approach false appearance of a wide-necked an-
greatly reduced the chances of missed eurysm can be created because of the 2. Molyneux AJ, Kerr RS, Yu LM, et al. In-
diagnosis. Meanwhile, clear display of limited resolution of 3D TOF MR angi- ternational subarachnoid aneurysm trial
(ISAT) of neurosurgical clipping versus
blood flow protrusion within arterial ography. This is also a probable reason
endovascular coiling in 2143 patients with
wall plaque and tortuous appearance for why neck width measurements at ruptured intracranial aneurysms: a ran-
of the initiation part of branches can 3D TOF MR angiography were mostly domised comparison of effects on survival,
mimic an aneurysm sac; infundibula greater than those at DSA in our study. dependency, seizures, rebleeding, sub-
can also be mistaken for aneurysms at Our study had limitations. This was groups, and aneurysm occlusion. Lancet
MR angiography if the vessel emerging a single-center experience, and recruit- 2005;366(9488):809–817.
at the apex of the infundibula is not ment was probably biased because in- 3. van Asch CJ, Velthuis BK, Rinkel GJ, et al.
seen (14). formed consent was obtained after non- Diagnostic yield and accuracy of CT angiog-
Data are still not available on the enhanced CT. Moreover, considering raphy, MR angiography, and digital subtrac-
tion angiography for detection of macrovas-
feasibility and safety of 3D TOF MR that a large proportion of patients who
cular causes of intracerebral haemorrhage:
angiography in cases of ruptured an- present with subarachnoid hemorrhage prospective, multicentre cohort study. BMJ
eurysms. Whether it can increase the will not be awake or will not be able to 2015;351:h5762.
risk for hemorrhage is a crucial issue to hold still to undergo good-quality MR
4. Lu L, Zhang LJ, Poon CS, et al. Digital sub-
be considered before its application. In angiography, CT angiography remains
traction CT angiography for detection of
our 3D TOF MR angiographic protocol, the first-line noninvasive examination intracranial aneurysms: comparison with
the imaging time was approximately 9 for this patient population. The risk of three-dimensional digital subtraction angi-
minutes; this is why we included only waiting for the MR imaging examination ography. Radiology 2012;262(2):605–612.
5. Nael K, Villablanca JP, Mossaz L, et al. 3-T 9. Suh SH, Cloft HJ, Huston J 3rd, Han KH, 13. Li MH, Chen SW, Li YD, et al. Prevalence of
contrast-enhanced MR angiography in eval- Kallmes DF. Interobserver variability of unruptured cerebral aneurysms in Chinese
uation of suspected intracranial aneurysm: aneurysm morphology: discrimination adults aged 35 to 75 years: a cross-sec-
comparison with MDCT angiography. AJR of the daughter sac. J Neurointerv Surg tional study. Ann Intern Med 2013;159(8):
Am J Roentgenol 2008;190(2):389–395. 2016;8(1):38–41. 514–521.
6. Li MH, Li YD, Tan HQ, et al. Contrast-free 10. Agresti A, Coull BA. Approximate is better 14. Lu H, Nagae-Poetscher LM, Golay X, Lin
MRA at 3.0 T for the detection of intra- than “exact” for interval estimation of bi- D, Pomper M, van Zijl PC. Routine clinical
cranial aneurysms. Neurology 2011;77(7): nomial proportions. Am Stat 1998;52(2): brain MRI sequences for use at 3.0 Tesla. J
667–676. 119–126. Magn Reson Imaging 2005;22(1):13–22.
7. Li MH, Li YD, Gu BX, et al. Accurate diag- 11. Ni W, Tian Y, Jiang H, et al. Preliminary 15. Chen YC, Sun ZK, Li MH, et al. The clini-
nosis of small cerebral aneurysms 5 mm in experience of 256-row multidetector com- cal value of MRA at 3.0 T for the diagnosis
diameter with 3.0-T MR angiography. Radi- puted tomographic angiography for detect- and therapeutic planning of patients with
ology 2014;271(2):553–560. ing cerebral aneurysms. J Comput Assist subarachnoid haemorrhage. Eur Radiol
Tomogr 2013;37(2):233–241. 2012;22(7):1404–1412.
8. Li MH, Cheng YS, Li YD, et al. Large-co-
hort comparison between three-dimensional 12. Chen W, Xing W, Peng Y, He Z, Wang C, 16. Cirillo M, Scomazzoni F, Cirillo L, et al.
time-of-flight magnetic resonance and rota- Wang Q. Cerebral aneurysms: accuracy of Comparison of 3D TOF-MRA and 3D CE-
tional digital subtraction angiographies in 320-detector row nonsubtracted and sub- MRA at 3T for imaging of intracranial
intracranial aneurysm detection. Stroke tracted volumetric CT angiography for diag- aneurysms. Eur J Radiol 2013;82(12):
2009;40(9):3127–3129. nosis. Radiology 2013;269(3):841–849. e853–e859.