Professional Documents
Culture Documents
Obesidad
Obesidad
Background and Purpose—The ASPECTS (Alberta Stroke Program Early CT Score) is a quantitate score that measures the
extent of early ischemic changes. Our aim was to investigate how measurement of ASPECTS using Hounsfield unit (HU)
values on initial noncontrast head computerized tomography (CT) correlates with the extent of final infarct on follow-up
imaging.
Methods—Cases of acute stroke from the middle cerebral artery M1 occlusion in which complete recanalization (TICI
[Thrombolysis in Cerebral Infarction] 3) was achieved were included for analysis. Using HU ratio (HU affected/HU
control hemisphere) and HU difference (HU control−HU affected hemisphere) values, ASPECTS was measured on initial
CT imaging and correlated with final ASPECTS at 24 hours. The study cohort consisted of 41 patients with acute stroke
from the M1 occlusion. The mean time from stroke symptoms onset to baseline head CT imaging was 264 minutes and
from CT to TICI 3 recanalization was 142 minutes.
Results—HU ratio within the 0.94 to 0.96 ranges showed the highest correlation coefficient and lowest mean and median
errors with the final ASPECTS. The difference of 2.0 HU between the 2 hemispheres demonstrated the higher correlation
coefficient (r=0.71; P<0.0001) and the lowest mean and median absolute errors (1.4 and 1, respectively).
Conclusions—We established a simple algorithm for rapid and accurate assessment of ASPECTS on baseline CT
imaging to predict the extent of final stroke in patients with emergent large vessel occlusion who undergo endovascular
revascularization. (Stroke. 2017;48:1574-1579. DOI: 10.1161/STROKEAHA.117.016745.)
Key Words: algorithms ◼ follow-up studies ◼ middle cerebral artery ◼ stroke ◼ thrombectomy
Downloaded from http://ahajournals.org by on September 10, 2019
Received January 16, 2017; final revision received March 21, 2017; accepted April 3, 2017.
From the Department of Neurosurgery, University of South Florida, Tampa (M.M., C.T.P.); Department of Neurosurgery, University at Buffalo, NY
(A.H.S.); and Department of Neurosurgery, Medical University of South Carolina, Charleston (A.S.T.).
Correspondence to Maxim Mokin, MD, PhD, Department of Neurosurgery, University of South Florida, 2 Tampa General Cir, 7th Floor, Tampa, FL
33606. E-mail maximmokin@gmail.com
© 2017 American Heart Association, Inc.
Stroke is available at http://stroke.ahajournals.org DOI: 10.1161/STROKEAHA.117.016745
1574
Mokin et al ASPECTS and Hounsfield Units 1575
as well as serve as a platform for development of automated 10 separate regions corresponding to the standard middle cerebral
ASPECTS software. artery territory ASPECTS areas were manually outlined. HU values
were recorded from the affected (stroke) and the contralateral (con-
trol) hemispheres within the 10 regions, as shown in Figure 2. The
Methods regions were outlined blinded to the knowledge of the extent of final
The study was approved by the local institutional review board. We stroke on follow-up imaging. The HU ratio (rHU) was calculated by
retrospectively reviewed cases of patients with acute ischemic stroke dividing the HU value of the affected side by the HU value of the
who were treated with endovascular therapy between July 2014 and contralateral side.
June 2016. Figure 1 shows the selection process used in our study. MRI fluid-attenuated inversion recovery was performed at ≈24
Cases that met the following criteria were included: (1) occlusion hours to measure the extent of final infarct. If MRI was not avail-
of the middle cerebral artery M1 segment; (2) complete recanali- able, noncontrast CT study repeated at 24 hours was used. ASPECTS
zation corresponding to TICI (thrombolysis in cerebral infarction) values on follow-up imaging was calculated by counting the presence
3 score at the end of thrombectomy; and (3) availability of base- or absence of final infarct within the standard 10 ASPECTS regions.
line noncontrast CT head and a follow-up noncontrast CT or MRI
brain at 24 hours. Cases with hemorrhagic transformation preclud-
ing accurate interpretation of the extent of final infarct or scans
Statistical Analysis
with significant motion artifacts were excluded from analysis. All The rHU ASPECTS (HU affected/HU control hemisphere) ranging
patients had CT angiography performed for confirmation of large from 0.90 to 0.99 and HU difference (HU control−HU affected hemi-
vessel occlusion. We excluded cases with a tandem separate occlu- sphere) were compared with final MRI ASPECTS. Pearson r correla-
sion at a more proximal location (such as cervical or internal carotid tion coefficient, mean absolute error, and median absolute error were
artery terminus occlusion) or occlusion of other territories, such as calculated for the total ASPECTS scores for each individual case.
the anterior cerebral artery. Intravenous thrombolysis was not an Sensitivity and specificity were analyzed for each ratio and difference
exclusion criterion. We typically selected patients for thrombectomy interval, as well as corresponding receiver operating characteristic
based on the ASPECTS cutoff value of 6. Cases with ASPECTS<6 curves. Statistical analyzes were performed using GraphPad Prism
were also treated with endovascular therapy on an individual case version 7 for Windows, GraphPad Software, La Jolla, CA. P<0.05
by case basis. Patients eligible for systemic thrombolysis received was considered statistically significant.
intravenous tPA (tissue-type plasminogen activator) first and then
were treated intra-arterial thrombectomy. There were no patients
in our cohort in whom intravenous thrombolysis alone resulted in Results
successful recanalization, which was confirmed angiographically Forty-one patients with acute stroke from the middle cere-
at the beginning of thrombectomy. Patients ineligible for systemic bral artery M1 occlusion who achieved complete recanaliza-
thrombolysis were taken for intra-arterial thrombectomy directly. All tion were included in this study. The baseline characteristics
thrombectomy procedures were performed by 2 operators using the
same approach to thrombectomy: direct aspiration first, with a sub- of these patients are summarized in Table 1. Mean time from
sequent use of a stent retriever, if aspiration alone was unsuccessful stroke onset to baseline head CT imaging was 264 minutes
Downloaded from http://ahajournals.org by on September 10, 2019
in achieving TICI 3 results. and from CT to TICI 3 recanalization was 142 minutes. For
the 24-hour follow-up imaging, brain MRI was performed in
Image Analysis 95% (39 of 41) of patients, and the remaining 2 patients had a
All patients underwent 5-mm noncontrast head CT imaging per- repeat head CT. Mean and median ASPECTS at 24 hours were
formed using Brilliance 64 PHILIPS scanner with the following stan- 6.7 and 8, respectively. Figure 2C shows the distribution of
dard parameters: tube voltage, 140 kV; tube current, 320 mA; and 24-hour ASPECTS within the cohort. The majority of patients
rotation speed, 0.75 seconds. Using baseline noncontrast CT images, (83%) had ASPECTS ≥5.
Table 2 shows the correlation of ASPECTS calculated
based on the ratio of HU (HU affected/HU control hemi-
sphere) on baseline CT imaging with final ASPECTS values
on the 24-hour follow-up imaging. rHU within the 0.94 to
0.96 ranges showed the highest correlation coefficient and
lowest mean and median errors with the final ASPECTS. In
a multivariate analysis, no significant effect of intravenous
tPA use, age, National Institutes of Health Stroke Scale,
or demographic factors on 24-hour ASPECTS was found
(P=0.195).
The correlations of ASPECTS calculated by subtracting
HU of the affected hemisphere from the contralateral (control)
side with final ASPECTS are shown in Table 2. HU differ-
ence within the range of 1.5 to 2.0 showed the highest cor-
relation while minimizing mean and median absolute value.
Specifically, absolute difference of 2.0 HU between the 2
hemispheres demonstrated the higher correlation coefficient
(r=0.71; P<0.0001) and the lowest mean and median absolute
errors (1.4 and 1, respectively).
The diagnostic sensitivity and specificity of ASPECTS
Figure 1. Flow diagram showing selection of patients for the based on rHU or HU difference on initial CT with final
study. IA indicates intra-arterial; ICA, internal carotid artery; M1,
middle cerebral artery M1 segment; and TICI, thrombolysis in ASPECTS at 24 hours according to different threshold values
cerebral infarction. are represented in Table 3. Receiver operating characteristic
1576 Stroke June 2017
Figure 2. An example of initial computerized tomography head with Hounsfield unit (HU) values calculated between the 2 hemispheres.
A, Ten ASPECTS (Alberta Stroke Program Early CT Score) regions are outlined manually. HU values representing each individual region
are shown at the bottom. Solid lines label the affected hemisphere and dotted lines label contralateral (control) hemisphere. B, Follow-up
magnetic resonance imaging fluid-attenuated inversion recovery repeated at 24 hours shows the extent of final stroke, ASPECTS=5. C,
Distribution of ASPECTS at 24 hours. C indicates caudate; I, insular ribbon; IC, internal capsule; L, lentiform; and M1–6, middle cerebral
artery cortical regions 1 to 6.
Downloaded from http://ahajournals.org by on September 10, 2019
Table 2. Correlation Between ASPECTS Based on HU Ratio Table 3. Sensitivity (%) and Specificity (%) Between
and HU Difference With Final ASPECTS at 24 Hours ASPECTS Based on HU Ratio or HU Difference With Final
ASPECTS at 24 Hours According to Different Threshold Values
Mean Mean Median
Pearson r (Median) Absolute Absolute Cutoff Sensitivity, % Specificity, %
(95% CI) P Value ASPECTS Error Error
HU ratio (HU affected/HU control hemisphere)
HU ratio (HU affected/HU control hemisphere)
<0.85 1.46 100
0.90 0.46 (0.18–0.68) 0.0011 4.2 (4) 2.7 2
<0.86 2.19 100
0.91 0.48 (0.20–0.69) 0.0007 5.2 (5) 2.4 2
<0.87 3.65 99.63
0.92 0.51 (0.24–0.71) 0.0003 6.0 (6) 2.2 1
<0.88 6.569 99.27
0.93 0.50 (0.22–0.70) 0.0005 6.9 (7) 1.8 1
<0.89 8.029 98.17
0.94 0.65 (0.43–0.80) <0.0001 7.5 (8) 1.6 1
<0.90 16.06 97.8
0.95 0.65 (0.42–0.80) <0.0001 8.1 (8) 1.4 1
<0.91 21.9 96.34
0.96 0.69 (0.49–0.82) <0.0001 8.5 (9) 1.3 2
<0.92 24.82 94.87
0.97 0.55 (0.29–0.73) 0.0001 8.8 (9) 1.7 2
<0.93 34.31 94.14
0.98 0.53 (0.26–0.72) 0.0002 9.0 (10) 2.0 3
<0.94 44.53 93.04
0.99 0.51 (0.24–0.71) 0.0003 9.3 (10) 2.7 4
<0.95 53.28 89.01
HU difference (HU control−HU affected)
<0.96 61.31 83.88
≥1 0.59 (0.35–0.76) <0.0001 5.9 (6) 1.5 2
<0.97 69.34 74.36
≥1.5 0.64 (0.42–0.79) <0.0001 7.0 (7) 1.3 2
<0.98 75.18 65.2
≥2 0.71 (0.52–0.84) <0.0001 7.9 (8) 1.4 1
<0.99 81.75 54.21
≥2.5 0.50 (0.23–0.70) 0.0008 8.6 (9) 1.9 1
HU difference (HU control−HU affected hemisphere)
≥3 0.49 (0.21–0.69) 0.0012 9.0 (9) 2.3 1
>1.0 69.85 73.26
≥3.5 0.42 (0.13–0.65) 0.0057 9.4 (10) 2.7 2
>1.5 58.82 84.98
ASPECTS indicates Alberta Stroke Program Early CT Score; CI, confidence
Downloaded from http://ahajournals.org by on September 10, 2019
Figure 3. Receiver operating characteristic analysis for prediction of final infarct on repeat imaging at 24 hours. A, Predicted ASPECTS
(Alberta Stroke Program Early CT Score) based on Hounsfield unit (HU) ratio thresholds. B, Predicted ASPECTS based on HU difference
thresholds. Both curves seem similarly, indicating comparable sensitivity, specificity, and accuracy of both algorithms. The 45° diagonal
line represents the line of no discrimination corresponding to the area under the curve (AUC) of 0.5.
for endovascular therapy, as well as serve as a foundation to Study (PICS) Investigators. Impact of pretreatment noncontrast CT
Alberta Stroke Program Early CT Score on clinical outcome after
develop automated software analysis for ASPECTS measure-
intra-arterial stroke therapy. Stroke. 2014;45:746–751. doi: 10.1161/
ment. The optimal rHU between the affected and contralat- STROKEAHA.113.004260.
eral hemispheres to predict final infarct on follow-up imaging 3. Yoo AJ, Berkhemer OA, Fransen PS, van den Berg LA, Beumer D,
was 0.94 to 0.96. Our findings need to be validated prospec- Lingsma HF, et al; MR CLEAN Investigators. Effect of baseline
Alberta Stroke Program Early CT Score on safety and efficacy of
tively by comparing such automated algorithm with human intra-arterial treatment: a subgroup analysis of a randomised phase 3
interpretation of ASPECTS. trial (MR CLEAN). Lancet Neurol. 2016;15:685–694. doi: 10.1016/
Downloaded from http://ahajournals.org by on September 10, 2019
S1474-4422(16)00124-1.
4. Powers WJ, Derdeyn CP, Biller J, Coffey CS, Hoh BL, Jauch EC, et
Disclosures al; American Heart Association Stroke Council. 2015 American Heart
Dr Siddiqui received grants from National Institutes of Health/National Association/American Stroke Association Focused Update of the 2013
Institute of Neurological Disorders and Stroke/National Institute of Guidelines for the Early Management of Patients With Acute Ischemic
Biomedical Imaging and Bioengineering and University at Buffalo Stroke Regarding Endovascular Treatment: a guideline for health-
(none related to this study); he received financial interests from care professionals from the American Heart Association/American
Hotspur, Intratech Medical, StimSox, Valor Medical, Blockade Stroke Association. Stroke. 2015;46:3020–3035. doi: 10.1161/STR.
Medical, and Lazarus Effect; he reports serving as a consultant 0000000000000074.
for Codman & Shurtleff, Inc, Concentric Medical, ev3/Covidien 5. Mainali S, Wahba M, Elijovich L. Detection of early ischemic changes in
Vascular Therapies, GuidePoint Global Consulting, Penumbra, noncontrast CT head improved with “Stroke Windows”. ISRN Neurosci.
Stryker, Pulsar Vascular, MicroVention, Lazarus Effect, and 2014;2014:654980. doi: 10.1155/2014/654980.
Blockade Medical; he reports serving on the speakers’ bureau for 6. Wardlaw JM, Farrall AJ, Perry D, von Kummer R, Mielke O, Moulin
Codman & Shurtleff, Inc, National Steering Committee–Penumbra T, et al; Acute Cerebral CT Evaluation of Stroke Study (ACCESS)
Study Group. Factors influencing the detection of early CT signs of
Inc’s 3D Separator Trial, Covidien’s SWIFT PRIME trial (Solitaire
cerebral ischemia: an internet-based, international multiobserver
With the Intention for Thrombectomy as Primary Endovascular
study. Stroke. 2007;38:1250–1256. doi: 10.1161/01.STR.0000259715.
Treatment), and MicroVention’s FRED trial (Pivotal Study of
53166.25.
the MicroVention Flow Re-Direction Endoluminal Device Stent 7. Demaerschalk BM, Vargas JE, Channer DD, Noble BN, Kiernan TE,
System in the Treatment of Intracranial Aneurysms); he is a mem- Gleason EA, et al. Smartphone teleradiology application is success-
ber of advisory boards for Codman & Shurtleff and Covidien fully incorporated into a telestroke network environment. Stroke.
Neurovascular; he received honoraria from Abbott Vascular, 2012;43:3098–3101. doi: 10.1161/STROKEAHA.112.669325.
Codman & Shurtleff, and Penumbra Inc. Dr Turk is a consultant 8. Sanossian N, Fu KA, Liebeskind DS, Starkman S, Hamilton S,
for Stryker, Codman, Penumbra, and Microvention and he received Villablanca JP, et al. Utilization of emergent neuroimaging for throm-
research grants (not related to this study) from Stryker, Codman, bolysis-eligible stroke patients. J Neuroimaging. 2017;27:59–64. doi:
Micorvention, Penumbra, and Covidien; he also received financial 10.1111/jon.12369.
interests from Medina Medical, Lazarus Effect, Pulsar Vascular, and 9. Hui FK, Obuchowski NA, John S, Toth G, Katzan I, Wisco D, et al.
Blockade Medical. The other authors report no conflicts. ASPECTS discrepancies between CT and MR imaging: analysis and
implications for triage protocols in acute ischemic stroke. J Neurointerv
Surg. 2017;9:240–243. doi: 10.1136/neurintsurg-2015-012188.
References 10. Herweh C, Ringleb PA, Rauch G, Gerry S, Behrens L, Möhlenbruch M, et
1. Barber PA, Demchuk AM, Zhang J, Buchan AM. Validity and reli- al. Performance of e-ASPECTS software in comparison to that of stroke
ability of a quantitative computed tomography score in predicting physicians on assessing CT scans of acute ischemic stroke patients. Int J
outcome of hyperacute stroke before thrombolytic therapy. ASPECTS Stroke. 2016;11:438–445. doi: 10.1177/1747493016632244.
Study Group. Alberta Stroke Programme Early CT Score. Lancet. 11. Gomolka RS, Chrzan RM, Urbanik A, Nowinski WL. A quantitative
2000;355:1670–1674. method using head noncontrast CT scans to detect hyperacute nonvisible
2. Yoo AJ, Zaidat OO, Chaudhry ZA, Berkhemer OA, González RG, ischemic changes in patients with stroke. J Neuroimaging. 2016;26:581–
Goyal M, et al; Penumbra Pivotal and Penumbra Imaging Collaborative 587. doi: 10.1111/jon.12363.
Mokin et al ASPECTS and Hounsfield Units 1579
12. Labeyrie MA, Turc G, Hess A, Hervo P, Mas JL, Meder JF, et al. vessel occlusion strokes. J Neurointerv Surg. 2017;9:442–444. doi:
Diffusion lesion reversal after thrombolysis: a MR correlate of early 10.1136/neurintsurg-2016-012331.
neurological improvement. Stroke. 2012;43:2986–2991. doi: 10.1161/ 15. Wintermark M, Flanders AE, Velthuis B, Meuli R, van Leeuwen M,
STROKEAHA.112.661009. Goldsher D, et al. Perfusion-CT assessment of infarct core and pen-
13. Soize S, Tisserand M, Charron S, Turc G, Ben Hassen W, Labeyrie MA, umbra: receiver operating characteristic curve analysis in 130 patients
et al. How sustained is 24-hour diffusion-weighted imaging lesion rever- suspected of acute hemispheric stroke. Stroke. 2006;37:979–985. doi:
sal? Serial magnetic resonance imaging in a patient cohort thrombolyzed 10.1161/01.STR.0000209238.61459.39.
within 4.5 hours of stroke onset. Stroke. 2015;46:704–710. doi: 10.1161/ 16. Eilaghi A, Brooks J, d’Esterre C, Zhang L, Swartz RH, Lee TY, et al.
STROKEAHA.114.008322. Reperfusion is a stronger predictor of good clinical outcome than recanali-
14. Mokin M, Gupta R, Guerrero WR, Rose DZ, Burgin WS, Sivakanthan zation in ischemic stroke. Radiology. 2013;269:240–248. doi: 10.1148/
S. ASPECTS decay during inter-facility transfer in patients with large radiol.13122327.
Downloaded from http://ahajournals.org by on September 10, 2019