Quantitative magnetic resonance angiography as an alternative imaging technique in the assessment of cerebral vasospasm after subarachnoid hemorrhage

Abstract

Introduction

The major mechanism of morbidity of delayed cerebral ischemia after subarachnoid hemorrhage (SAH) is considered to be severe vasospasm. Quantitative MRA (QMRA) provides direct measurements of vessel-specific volumetric blood flow and may permit a clinically relevant assessment of the risk of ischemia secondary to cerebral vasospasm.

Purpose

To evaluate the utility of QMRA as an alternative imaging technique for the assessment of cerebral vasospasm after SAH.

Methods

QMRA volumetric flow rates of the anterior cerebral artery (ACA), middle cerebral artery (MCA), and posterior cerebral artery (PCA) were compared with vessel diameters on catheter-based angiography. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of QMRA for detecting cerebral vasospasm was determined by receiver-operating characteristic curves. Spearman correlation coefficients were calculated for QMRA flow versus angiographic vessel diameter.

Results

Sixty-six vessels (10 patients) were evaluated with QMRA and catheter-based angiography. The median percent QMRA flow of all vessels with angiographic vasospasm (55.0%, IQR 34.3–71.6%) was significantly lower than the median percent QMRA flow of vessels without vasospasm (91.4%, IQR 81.4–100.4%) (p < 0.001). Angiographic vasospasm reduced QMRA-assessed flow by 23 ± 5 (p = 0.018), 95 ± 12 (p = 0.042), and 16 ± 4 mL/min (p = 0.153) in the ACA, MCA, and PCA, respectively, compared to vessels without angiographic vasospasm. The sensitivity, specificity, PPV, and NPV of QMRA for the discrimination of cerebral vasospasm was 84%, 72%, 84%, and 72%, respectively, for angiographic vasospasm >25% and 91%, 60%, 87%, and 69%, respectively, for angiographic vasospasm >50%. The Spearman correlation indicated a significant association between QMRA flows and vessel diameters (r_s = 0.71, p < 0.001).

Conclusion

Reduction in QMRA flow correlates with angiographic vessel narrowing and may be useful as a non-invasive imaging modality for the detection of cerebral vasospasm after SAH.

Introduction

Aneurysmal subarachnoid hemorrhage (SAH) is a devastating condition often complicated by delayed cerebral ischemia (DCI) in the days following the initial event. ¹ After hemorrhage, DCI afflicts nearly 40% of patients and represents the leading treatable predictor of long-term disability. ² Despite the complex mechanisms of DCI-related morbidity, many clinicians consider cerebral vasospasm a surrogate marker for DCI, and its early detection is the standard of care in many institutions. ³ Formal catheter-based cerebral angiography remains the gold standard to diagnose cerebral vasospasm, but due to its invasive nature, more convenient imaging modalities, such as CT angiography (CTA) and transcranial Doppler (TCD), are commonly employed for screening and diagnosis. ⁴ However, these less invasive modalities have inherent limitations and may fail to identify hemodynamic compromise in some patients.^5,6 For instance, CTA only provides anatomical data of cerebral vessels and TCD, although frequently applied in clinical practice, may not be reliable due to poor acoustic windows in some individuals.

Quantitative magnetic resonance angiography (QMRA) is a non-invasive technique to evaluate the volumetric blood flow rate of individual intracranial vessels.^7,8 It provides clinicians with direct quantification of blood flow and has been shown to be effective at distinguishing patients at risk of cerebral ischemia secondary to hemodynamic compromise in various cerebrovascular disorders.^9,10 QMRA may permit a more clinically relevant assessment of ischemia due to cerebral vasospasm and reduce the reliance on surrogate measures of cerebral blood flow, such as blood flow velocity and anatomic degree of vessel stenosis. To date, there are no studies that have investigated the ability of QMRA to detect cerebral vasospasm and evaluate its utility to provide an alternative non-invasive screening technique in those patients in which traditional methods, such as CTA and TCD, are limited. Thus, we conducted this single-center study to evaluate the utility of QMRA in the assessment of cerebral vasospasm after SAH.

Methods

Study population

We performed a retrospective analysis of a prospectively maintained database of all patients admitted with SAH who underwent a QMRA between post-bleed day 0 and post-bleed day 21 at a single institution. Between August 2017 and October 2021, 35 patients received a QMRA as a routine screening study for the evaluation of cerebral vasospasm. Of these, ten patients underwent catheter-based cerebral angiography within 24 h of QMRA and were included in this analysis. Demographic, clinical, and imaging data were collected from the electronic medical record. This study was approved by the local institutional review board (IRB# 20-0462). The requirement for written informed consent was waived.

Quantitative magnetic resonance angiography

The technique that allows for quantification of blood flow in intracranial and extracranial vessels using QMRA is previously described. ⁷ Images are acquired using cardiac-gated phase contrast MRA at a voxel that is placed perpendicular to the axis of the vessel of interest on time-of-flight MR angiography. The process of placing the region of interest voxel in optimal locations was supervised by a blinded senior neuroradiologist experienced in QMRA post-processing. Standard sites of measurement include 13 vessels (right vertebral artery, left vertebral artery, basilar artery, right posterior cerebral artery, left posterior cerebral artery, right internal carotid artery, left internal carotid artery, right anterior cerebral artery – A1 segment, right anterior cerebral artery – A2 segment, left anterior cerebral artery – A1 segment, left anterior cerebral artery – A2 segment, right middle cerebral artery, and left middle cerebral artery). Measurements of volumetric flow rate (mL/min) at the region of interest are obtained on a separate workstation using Non-invasive Optimal Vessel Analysis (NOVA, VasSol Inc.), a commercially available software. Age-adjusted normative ranges of specific vessel blood flows are derived from QMRA investigations of healthy subjects and are provided by the NOVA software. To minimize the variability of flow measurements introduced by anatomic variations of the circle of Willis and the effect of distal vasospasm on proximal vessel flows,^11,12 only the QMRA flows rates of A2 segment of the anterior cerebral artery (ACA), M1 segment of the middle cerebral artery (MCA), and P2 segment of the posterior cerebral artery (PCA) were included in the data analysis. Cerebral blood flow measurements were obtained during the “spasm period” (post-bleed day 5–14) for each vessel. A portion of patients underwent QMRA during the “baseline period” (post-bleed day 0–1) as well as the “spasm period” and are also included in the analyses. A total of 66 vessels were available for review.

Clinical management

All patients were managed according to a hospital-specific protocol for aneurysmal SAH. A standardized, tiered approach to DCI is utilized at our institution beginning with blood pressure augmentation and progressing towards intra-arterial vasodilator administration for rescue therapy. Patients received 6-vessel diagnostic cerebral angiography within 24 h of hospital admission and again within 24 h after their QMRA study. If patients were found to have low QMRA vessel flow, their clinical treatment was not altered or modified until catheter-based angiography confirmed the presence or absence of cerebral vasospasm. Angiographic vasospasm was defined as a reduction of the diameter of the vessel by greater than 25% between the baseline cerebral angiogram and the spasm period cerebral angiogram. This cutoff value was determined by experimental data in canine coronary flow reserve, which begins to decrease with stenosis of 30% of arterial diameter. ¹³ Measurements of baseline and spasm period vessel diameters were performed by an interventional neurologist and neurosurgeon blinded to the clinical data. Measurements of stenosis were performed at the site of greatest stenosis of a particular segment (A2, M1, or P2). Presence of radiographic evidence of infarction (MRI or CT) before discharge was recorded and long-term follow-up data was collected.

Statistical analysis

Statistical analyses were performed using R statistical software. Values are expressed as mean ± SEM or median with IQR as denoted. Data was assessed for normality using the Shapiro-Wilk test. Differences between groups were calculated using student's t-test or Wilcoxon signed-rank test for parametric and non-parametric data, respectively. Comparisons of QMRA flow between subgroups were assessed using contingency tables. To determine the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of QMRA for the detection of cerebral vasospasm, a formal sample size calculation was performed. Assuming an alpha level of 0.05, power of 0.80, and absolute precision level of 0.1, the sample size for diagnostic accuracy analyses was determined to be 56. Receiver operator characteristic curves were generated using two stenosis-specific models: angiographic vasospasm greater than 25% and angiographic vasospasm greater than 50%. The QMRA flows at which sensitivity and specificity were optimized were determined by the Youden Index method. Spearman correlation coefficients were calculated for QMRA flow versus angiographic vessel diameter and change in QMRA from baseline versus change in angiographic vessel diameter from baseline for those patients in which baseline QMRA data was available. In addition, a Bland-Altman agreement analysis was performed. The statistical significance level was set at 0.05.

Results

Patients

Ten patients underwent 11 QMRA examinations of 66 intracranial vessels during the spasm period. Six individuals of this cohort also underwent QMRA examinations during the baseline period. The median age of patients was 59 years (IQR 51.5–69.0), and was comprised of 90% females (9/10). The majority of patients were admitted with Hunt and Hess grade 1 or 2 (90%) and Fisher grade 3 or 4 (100%). All patients underwent balloon-assisted endovascular coil embolization of their ruptured aneurysm. No patient experienced significant intracranial hemorrhage or infarct after their target aneurysm had been secured. Four patients had ACOM artery aneurysms, 3 patients had MCA aneurysms, 1 patient had a PCOM artery aneurysm, 1 patient had an AICA-PICA aneurysm, and no aneurysm was identified in 1 patient. All patients received intra-arterial vasodilator therapy for symptomatic cerebral vasospasm during their clinical course, and 1 patient required repeat therapy for persistent cerebral vasospasm. The median time to spasm period QMRA from index hemorrhage was 7.0 days (IQR 6–8 days). The median time until catheter-based angiography after QMRA examination was 22:07 h (IQR 5:21–23:46 h). Of the vessels evaluated, 41.7% were found to have angiographic vasospasm. Table 1 summarizes demographic information and baseline characteristics of the cohort. By the time of discharge, one patient (10%) demonstrated scattered, punctate infarcts within the right frontal lobe without neurological deficit, one patient (10%) died from progressive infarction from refractory vasospasm, and the remainder (80%) did not show any radiographic evidence of infarction. The median time of long-term follow up was 5.0 months (IQR 4.1–6.2 months). At the time of last follow-up, all patients had a modified Rankin score 0–1.

Table 1.

Baseline characteristics of cohort.

Category	Number	Proportion (%)
Number of patients	10	-
Vessels evaluated for vasospasm	66	-
Gender
Male	1/10	10
Female	9/10	90
Age
>51 years	7/10	70
<51 years	3/10	30
Hunt and Hess grade
1–2	9/10	90
3–5	1/10	10
Fisher grade
1–2	0/10	0
3–4	10/10	100
Endovascular coiling	10/10	100
Surgical clipping	0/10	0
Intra-arterial vasodilator treatment	10/10	100
Repeated intra-arterial vasodilator treatment	3/10	30
Aneurysm location
ACOM	4/10	40
MCA	2/10	20
PCOM	1/10	10
AICA/PICA	1/10	10
No aneurysm identified	2/10	20
Vessels with angiographic vasospasm
ACA	11/20	55
MCA	8/20	40
PCA	6/20	30
All	25/60	41.7

ACOM, anterior communicating artery; MCA, middle cerebral artery; PCOM, posterior communicating artery; AICA, anterior inferior cerebellar artery; PICA, posterior inferior cerebellar artery; ACA, anterior cerebral artery; PCA, posterior cerebral artery.

QMRA flows

Overall, the percent of QMRA flow of previously published average normative values was statistically significantly decreased in vessels with angiographic vasospasm (55.0% (IQR 34.3–71.6%) vs. 91.5% (IQR 81.4–100.4%), p < 0.001) (Figure 1). The ACA was found to have angiographic vasospasm most frequently (n = 11), followed by the MCA (n = 8) and the PCA (n = 6). In vessel-specific subgroup analyses, the QMRA flow values of ACA and MCA vessels with angiographic vasospasm were below established average normative flow values. Specifically, the QMRA flows of spastic ACA and MCA vessels were 31 ± 4 mL/min (60.1% reduction compared to average normative value) and 77 ± 9 mL/min (54.1% reduction compared to average normative value), respectively. The QMRA flow of PCA vessels with angiographic vasospasm was 73 ± 11 mL/min (0.23% reduction compared to average normative value) and within the established normative range of PCA flow rates.

Figure 1.

Percent QMRA flow of all vessels evaluated in comparison to average normalized values. The percent QMRA flow of all vessels with angiographic vasospasm was significantly lower than those vessels without angiographic vasospasm (p < 0.001). VS, vasospasm.

Subgroup analysis

A subgroup analysis was performed on 36 vessels with baseline QMRA flows. Five ACA vessels, 3 MCA vessels, and 2 PCA vessels demonstrated >25% stenosis on angiography performed during the spasm period, and 7 ACA vessels, 9 MCA vessels, and 10 PCA vessels did not demonstrate angiographic vasospasm. The difference in QMRA flow between the baseline and spasm period was −46.4 mL/min in vessels with angiographically confirmed vasospasm (96.7 vs. 50.3 mL/min, p = 0.002), while the QMRA flow in vessels without angiographic vasospasm changed by −3.5 mL/min (82.8 vs. 79.3 mL/min, p = 0.52). Specifically, QMRA flows were reduced by 23 ± 5 mL/min (p = 0.018), 95 ± 12 mL/min (p = 0.042), and 16 ± 4 mL/min (p = 0.153) in the ACA, MCA, and PCA, respectively, between the baseline QMRA and spasm period QMRA. These results are depicted in Table 2 and Figure 2. Overall, the QMRA flow decreased in 100% of vessels (10/10) with angiographic vasospasm and 16% of vessels (4/25) without angiographic vasospasm (Figure 3).

Table 2.

Flow values of baseline and spasm period QMRA of ACA, MCA, and PCA vessels.

Category	Baselinea (n = 36)	Spasm Perioda (n = 66)	Differencea	p-value
No angiographic VS
ACA (n = 7)	53 ± 7	54 ± 8	1 ± 6	0.916
MCA (n = 9)	126 ± 13	117 ± 3	−9 ± 15	0.445
PCA (n = 10)	63 ± 5	69 ± 4	6 ± 4	0.531
Angiographic VS
ACA (n = 5)	54 ± 7	31 ± 4	−23 ± 5	0.018
MCA (n = 3)	172 ± 10	77 ± 9	−95 ± 12	0.042
PCA (n = 2)	89 ± 1	73 ± 11	−16 ± 4	0.153

^aData is presented as mean ± SEM mL/min. ACA, anterior cerebral artery; MCA, middle cerebral artery; PCA, posterior cerebral artery; VS, vasospasm.

Figure 2.

Vessel-specific QMRA flow values (mL/min) of ACA, MCA, and PCA with and without angiographic vasospasm. (A-C) QMRA flows are significantly reduced with angiographic vasospasm of the ACA and MCA, however no change in QMRA flow was seen in PCA vessels with angiographic vasospasm. No significant change in QMRA flow was observed in vessels without angiographic vasospasm. (D) In comparison to baseline QMRA flow, significant reductions in flow were observed in the ACA and MCA with angiographic vasospasm, but not in the PCA. Data is displayed as mean ± SEM. ACA, anterior cerebral artery; MCA, middle cerebral artery; PCA, posterior cerebral artery; VS, vasospasm.

Figure 3.

Receiver operator characteristic curves. Two conditions based on degree of angiographic vasospasm were modeled. The AUC for angiographic vasospasm >25% was 0.8325. The AUC for angiographic vasospasm >50% was 0.8267. AUC, area under the curve; VS, vasospasm.

Diagnostic accuracy of QMRA

Two receiver operator characteristic curves including all vessels were modeled based on the cutoff points of angiographic vasospasm greater than 25% and angiographic vasospasm greater than 50%. The overall performance (AUC) of the two models established by the cutoff points was 0.8325 and 0.8267 for angiographic vasospasm >25% and angiographic vasospasm >50%, respectively (Figure 4). The Youden Index method was used to optimize the performance of the ROC curves. Using this method, the sensitivity, specificity, PPV, and NPV of QMRA for the discrimination of cerebral vasospasm was 84%, 72%, 84%, and 72%, respectively, for angiographic vasospasm >25% and 91%, 60%, 87%, and 69%, respectively, for angiographic vasospasm >50% (Table 3).

Figure 4.

Change in QMRA blood flow in comparison to individualized baseline values. All vessels with angiographic vasospasm demonstrated a decrease in QMRA flow between baseline QMRA and spasm period QMRA examinations. The QMRA blood flow in vessels without angiographic vasospasm did not significantly change between the two time points. VS, vasospasm.

Table 3.

Sensitivity, specificity, positive predictive value, and negative predicative value of QMRA for the detection of cerebral vasospasm.

	Sensitivity	Specificity	PPV	NPV
Angiographic VS >25%	84%	72%	84%	72%
Angiographic VS >50%	91%	60%	87%	69%

PPV, positive predictive value; NPV, negative predictive value; VS, vasospasm.

Correlation of QMRA with vasospasm

The result of the Spearman correlation indicated a significant association between QMRA flows and vessel diameters (r_s = 0.71, p < 0.001, n = 66) (Figure 5(a)). This correlation was further increased when individual baseline QMRA flows were used for analysis (r_s = 0.83, p < 0.001, n = 36) (Figure 5(b)). Bland-Altman agreement analysis was performed with the subset of patients with individualized baseline QMRA flows and is shown in Figure 6. The estimated bias was 0.03, and the limits of agreement were −0.31 and 0.39.

Figure 5.

Spearman correlation. (A) Absolute values of QMRA flow (mL/min) versus vessel diameter (mm) demonstrates a significant positive correlation (r_s = 0.71, p < 0.001). (B) Percent change in QMRA flow and percent change in vessel diameter were obtained from baseline values available for 36 vessels. Using this method, QMRA flow and vessel size demonstrates a stronger positive correlation (r_s = 0.83, p < 0.001).

Figure 6.

Bland-Altman plot comparing change in QMRA flows and change in vessel diameters on angiography demonstrated moderate agreement between the two modalities across lower and higher values. Solid line: estimated bias. Dashed lines: Bland-Altman limits of agreement.

Discussion

Quantitative MRA has been shown to be useful in the assessment of a variety of flow-limiting cerebrovascular conditions^14–22 and has been validated as a primary tool to assess stroke risk in vertebrobasilar insufficiency. ²³ To the best of our knowledge, our study is the first to report the use of QMRA in cerebral vasospasm after SAH. Our results indicate that angiographic vasospasm was significantly associated with reduced QMRA-measured blood flow in the ACA, MCA, and PCA. Further, we found that reduced QMRA flow predicts cerebral vasospasm with high sensitivity, high PPV, and fair NPV, suggesting that QMRA may be a feasible surveillance technique in patients at risk of developing DCI.

Cerebral vasospasm is strongly associated with neurologic deterioration after SAH and its early diagnosis and treatment remains a key component of DCI management.^24,25 As such, the development of effective non-invasive imaging modalities to identify arterial narrowing and appropriately stratify the risk of infarction after SAH has been of significant interest. Since the first reported use of transcranial Doppler in SAH, ²⁶ detection of vasospasm with transcranial Doppler has gained popularity and is currently recommended by the American Heart Association/American Stroke Association (Class IIA/Level B evidence). ²⁷ However, the transcranial Doppler technique is highly operator dependent and can be limited by inadequate acoustic windows, which occurs in about 8% of patients. ⁶ In cases where transcranial Doppler examinations are unreliable and patients are challenging to monitor clinically, CTA and CT perfusion (CTP) may be performed, but the poor agreement between interpreters, obscuration of intracranial vessels by metal streak artifacts, and the additional contrast-dye and radiation dose to patients limits its widespread use.^5,28 In particular, CTP, which has gained rapid popularity for its excellent assessment of tissue-level hemodynamics, provides limited evaluation of the posterior fossa vascular territories. Alternative techniques, such as SPECT or perfusion-weighted MRI, are either time-consuming, burdensome, subject patients to radiation exposure, or importantly, do not provide direct quantification of blood flow. ²⁹ Further, changes seen in microcirculatory hemodynamics in perfusion-related imaging may not necessarily reflect proximal or distal vasospasm but rather other sequelae of DCI, such as neurovascular decoupling or neuroinflammation, and may not respond to typical measures against vasospasm-induced DCI. QMRA is a promising technique for the detection of flow-related compromise, and the direct quantification of individual vessel blood flow by QMRA may provide an easily interpretable assessment of the risk of ischemia in patients with cerebral vasospasm. The utility of QMRA may be maximized in those patients who lack acoustic windows for serial TCD examinations and cannot tolerate repeated contrast boluses for CTA/CTP. It is in these situations that QMRA may potentially provide the greatest clinical benefit.

The definition of “hemodynamically significant” stenosis is unclear in the cerebrovascular literature. While popular assessments of stroke risk rely on linear measurements of arterial narrowing, such as arterial diameter, to define flow compromised states, most investigators agree that the relationship between arterial stenosis and flow reduction is unpredictable and non-linear. ³⁰ There exists a subset of patients that experience symptoms of hypoperfusion at low degrees of arterial stenosis.^31,32 Thus, a solely “anatomic stenosis-based” approach to states of hypoperfusion ignores the hemodynamic relevance of some luminal stenoses. For example, the effects of tandem stenoses are additive and their impact is poorly assessed on most current screening modalities for cerebral vasospasm. This begs the question, are established thresholds of critical stenosis a reliable indicator of regional cerebral hypoperfusion in multifocal disease such as SAH-induced vasospasm? While we did not specifically address this question in the current study, QMRA has the potential to offer a more functional and direct measure of hemodynamic compromise in SAH. Given that QMRA is able to evaluate various physiologic factors besides volumetric flow, such as resistivity and pulsatility index, it may offer a more accurate assessment of microvascular spasm and resistance of the distal capillary bed than other assessments of solely large arteries. In fact, in our cohort, there were 3 patients in which the quantitative vessel blood flow was below the lower limit of normative values, yet only mild (<25%) stenosis was observed on angiography. Considering that existing literature on cerebrovascular hemodynamics suggests a reduction of distal cerebral blood flow of 20–25% can be associated with cerebral ischemia,^33–35 our findings highlight that physiologically significant hypoperfusion may be present in patients with DCI that is not anatomically apparent. This may be especially true in cases where distal vasospasm dominates the clinical picture.

Overall, the results of this pilot study demonstrate that QMRA provides accurate vessel-specific discrimination of cerebral vasospasm in a cohort of patients with SAH. Additionally, QMRA may be particularly helpful in patients without acoustic windows for transcranial Dopplers or poor tolerance for serial CT angiograms and for whom the risk of catheter-based angiography is to be minimized. The QMRA protocol can be abbreviated or individualized to only include flow measurements of distal arteries, like in this report, or a single hemisphere, and additional sequences, such as diffusion-weighted imaging, can be included for greater clarification. At our institution, a standard 13-vessel QMRA is acquired at baseline if feasible and subsequent QMRA examinations include only vessels of interest, dramatically reducing the MRI scan time.

Our study has some limitations. First, the findings are limited by the small sample size and retrospective nature of the analysis. However, this is the largest study to date to evaluate the use of QMRA in cerebral vasospasm. Second, relative flow reductions were compared to normative flow ranges available from the proprietary software NOVA, but the cutoff value that defines abnormality in cerebral vasospasm is not defined. It is known that age and anatomical variants can modify blood flow values on QMRA, but conditions like blood pressure augmentation or cardiac dysfunction can also affect QMRA flows.^11,30,36 The effects of these conditions on QMRA flow after SAH need to be further investigated. For example, the variability of flow values introduced by anatomical variants may be minimized by assessments of hemispheric blood flow, total cerebral flow, or individualized baseline vessel flows as we have demonstrated in a subset of our cohort. Future studies should aim to obtain baseline QMRA of subjects to increase diagnostic accuracy. Third, this study is limited by the fact that the majority of patients were admitted with low Hunt and Hess grades. While this improved the feasibility of obtaining QMRA examinations, it is important to note that this approach may limited by the difficulty of obtaining MR examinations in critically ill patients. MR incompatible invasive monitoring devices (PbO2 sensors, ICP monitors, etc.) are contraindications to QMRA examination, however the proportion of patients that require these is a small fraction of all patients with SAH. As the use of MRI becomes more important in neurocritical care, newer devices that are MR conditional or MR safe are becoming available, such as the Neurovent® ICP monitor (Raumedic AG, Münchberg, Germany), opening the possibilities for serial MR examinations or portable bedside MRI in SAH patients. At our institution, MR compatible devices are implanted as frequently as possible when required and the process of transporting ICU patients to the MRI suite is optimized. Lastly, the optimal timing of QMRA examination after SAH is not clear. The question of whether QMRA offers the discriminatory power and diagnostic accuracy to identify vasospasm in individuals with less severe disease remains. Future research focused on early reductions in cerebral blood flow before they are clinically apparent is warranted.

Conclusion

We show that reduction in QMRA blood flow is an indicator of angiographic vessel narrowing and may offer an alternative assessment of the risk of ischemia in SAH-induced vasospasm. Thus, QMRA represents a feasible, non-invasive imaging modality for the detection of cerebral vasospasm in patients with SAH.