Sage Journals: Discover world-class research

Abstract

Introduction

Determining the cause of spontaneous (non-traumatic) intracerebral haemorrhage (ICH) is critical to guide treatment and prognosis. We investigated whether small vessel disease (SVD) in addition to clinical and other radiological findings on acute neuroimaging predicts a low risk of a macrovascular cause (e.g. an arterio-venous malformation, aneurysm or dural arteriovenous fistula).

Patients and methods

We identified patients with acute spontaneous ICH who underwent acute non-contrast CT, CT angiography (CTA) and intra-arterial digital subtraction angiography (IADSA) at our institution from January 2010 to April 2014. Logistic regression including CTA result, SVD, age, pre-ICH hypertension and ICH location was used to derive a prediction model, validated using bootstrapping.

Results

173 patients (46% female, median age 49) of whom 78 had a macrovascular cause on IADSA were included. Predictors of a macrovascular cause were: abnormal CTA (OR 67.4; p < 0.001); absence of SVD (OR 5.0; p = 0.019); and absence of pre-ICH hypertension (OR 3.4; p = 0.05). In our internally derived prediction model, the combination of CTA, SVD and pre-ICH hypertension predicted the likelihood of an underlying macrovascular cause (optimism-adjusted ROC area 0.919). Patients with negative CTA, SVD and pre-ICH hypertension have a low likelihood of an underlying macrovascular cause (1.8%).

Discussion and conclusion

A combination of CTA, SVD and pre-ICH hypertension predict the likelihood of finding a macrovascular cause in patients with acute spontaneous ICH, allowing informed decisions regarding the likely benefit and risk of IADSA.

Keywords

CT angiography small vessel disease intracerebral haemorrhage intra-arterial digital subtraction angiogram

Introduction

Intracerebral haemorrhage (ICH) is devastating: mortality is 54% at 1 year,¹ while only 12–39% of survivors recover to independence.² Most cases (77–88%) of spontaneous (non-traumatic) intracerebral haemorrhage are termed ‘primary,’³ with a presumption that they are caused by small vessel disease (SVD).⁴ However, it is critical to exclude ‘secondary’ structural causes of ICH, including macrovascular causes (e.g. arterio-venous malformations, aneurysms and dural arteriovenous fistulae), which can be treated.

The ‘reference standard’ for detection of a macrovascular cause is intra-arterial digital subtraction angiography (IADSA) or neurosurgery. IADSA is invasive, requires skilled operators and is associated with a small but appreciable mortality and morbidity, especially in acute ICH.⁵ Selecting which ICH patients have a sufficiently high likelihood of a macrovascular cause to recommend IADSA is a common and important clinical question. Current practice varies widely:⁶ typically, the presence of pre-ICH hypertension, deep location of ICH and age are used as indicators of SVD to select patients unlikely to require IADSA,⁶ but the evidence supporting this is scant and conflicting.^7–9 A scoring system incorporating pre-ICH hypertension and age (among other factors) to identify patients at risk of a macrovascular cause has been previously validated, but did not include imaging markers of SVD and only had moderate discrimination outside of the United States.^10,11 SVD can be directly identified on brain imaging by leukoaraoisis and lacunar infarction, even on plain CT. We therefore hypothesised that, in acute ICH patients, visualisation of SVD will predict a low yield of a macrovascular cause; furthermore, we aimed to develop and internally validate an algorithm to help clinicians identify patients at high risk of having a macrovascular cause underlying acute spontaneous ICH.

Patient and methods

Patient selection

We retrospectively reviewed all consecutive patients who underwent IADSA (the reference standard for detecting a macrovascular cause of ICH) for the investigation of ICH at the National Hospital for Neurology and Neurosurgery, Queen Square, from January 2010 to April 2014. Inclusion criteria were: acute, symptomatic, non-traumatic (spontaneous) ICH, with availability of diagnostic quality CTA, non-contrast CT and an IADSA. Patients with a primary diagnosis of subarachnoid haemorrhage (SAH) or subdural haemorrhage were excluded. In our centre, all patients presenting to our stroke unit with acute ICH have acute non-contrast CT and CTA unless there is a contra-indication. The need for an IADSA is decided during a weekly multidisciplinary vascular neuroradiological meeting where the patient’s age, ICH location and history (including hypertension) are taken into account, as per standard clinical practice. We did not exclude patients on the basis of age. We recorded whether the patient had a suspicion of a macrovascular cause on the non-invasive imaging prior to the IADSA to explore partial verification bias.

Data collection

Variables collected were: age (dichotimised into aged 45 years or over vs. under 45, in keeping with current clinical practice and previous studies^8,12); pre ICH hypertension (defined by previously documented HTN for which either lifestyle advice or antihypertensive medication had been provided); and location of the ICH (cerebellar, intraventricular (pure), lobar and deep perforator territory (brainstem and basal ganglia)).

Image analysis

A trained vascular neuroradiology fellow (AO) reviewed the acute CTA, blinded to the IADSA result, for the presence of any structural macrovascular cause. This CTA evaluation was then compared to the final clinical report (by an accredited consultant vascular neuroradiologist). Disagreement was reviewed by a professor of vascular neuroradiology (HRJ) and a consensus decision was reached. A ‘negative’ CTA was defined as showing no indication of a macrovascular cause. A ‘positive’ CTA was defined by the final neuroradiology assessment as being suspicious for a macrovascular cause (graded as either having a ‘possible’ or ‘definite’ underlying macrovascular cause). A clinical research fellow (DW) trained in SVD rating rated the non-contrast CT for SVD using the simplified Fazekas scale¹³ and the presence of lacunae, blinded to the IADSA result. SVD on CT was dichotomised into moderate-severe (Fazekas grade ≥2 in either periventricular or deep white matter distribution) and/or the presence of any lacunar infarction vs. mild/none SVD and no lacunar infarctions. SVD was rated in the hemisphere contralateral to the symptomatic ICH to avoid misclassification of peri-haematomal oedema. Haematoma location was classified using a recently published rating instrument.¹⁴ The IADSA was reported by a certified consultant vascular neuroradiologist. The routine IADSA protocol for investigation of intracerebral haemorrhage at our institution includes selective catheterisation of the internal carotid, external carotid and vertebral arteries, with angiographic runs of each of these vessels in at least two projections and followed through to the venous phase.

Statistical analysis

Univariable odds ratios and the sensitivity and specificity (presented as ROC area under the curve) were undertaken for the following variables: age, ICH location, HTN, CTA result and SVD identification on CT against the reference standard of IADSA detected macrovascular causes.

We fitted a logistic regression model to generate risk coefficients for macrovascular causes. We checked the fit of the model by comparing the models prediction to the observed outcomes. This model was then internally validated using bootstrap validation with 1000 samples. Discrimination was quantified using the ROC area and calibration was assessed using the Cox–Miller calibration slope.

All analyses were performed using STATA 12.0 (StataCorp LP, TX).

The study was conducted and reported according to reference standards described in the TRIPOD guidelines (https://www.equator-network.org/reporting-guidelines/tripod-statement/)

Ethical approval

The study was approved by the Clinical Governance Committee of the National Hospital and the UCL Institute of Neurology and National Hospital Joint Research Ethics Committee.

Results

We identified 204 patients with acute spontaneous ICH who had an IADSA. The median age was 49 (range 18–86, IQR 40–59)) and 54% were males. After quality assurance, we included 173 patients with diagnostic quality CTA in the study (Figure 1). Within this group there were 78 IADSA-defined macrovascular causes (68 AVMs, 7 dural fistulas, 2 aneurysms and 1 carotid-cavernous fistula). The median CTA to IADSA time was 2 days (IQR 1–11). Three vascular malformations were only detected on repeat IADSA; of these patients the median CT-repeat DSA was 264 days (IQR 78–314). ICH locations in the final cohort were: lobar 83, deep perforator territory 60 (basal ganglia, thalamus, caudate, brainstem), Cerebellar 21, pure intraventricular: 9 (Table 1).

Figure 1.

Flow chart for study.

Table 1.

Characteristics of the patients in the study.

Variable	Entire cohort (n = 173)
Age, years, median (IQR) [Range]	49 (40–59) [18–86]
Gender, Female n(%)	80 (46)
Pre ICH HTN n(%)	52 (30)
CTA to IADSA, Days, median (IQR)	2 (1–11)
Abnormal CTA, n(%)	71 (41)
Confluent leukoaraiosis on CT, n(%)	41 (24)
Lacunar infarcts on CT, n(%)	13 (8)
Any SVD on CT, n(%)	47 (27)
Location of ICH: Cerebellar, n(%)	21 (12)
Location of ICH: Deep perforator territory, n(%)^a	60 (35)
Location of ICH: Pure intraventricular, n(%)	9 (5)
Location of ICH: Lobar, n(%)	83 (48)

HTN: hypertension; ICH: intracerebral haemorrhage; CTA: computed tomography angiography.

Brainstem, basal ganglia, thalamus and caudate nucleus.

CTA identified 85 cases with a definite or possible macrovascular cause. 67/67 cases identified as ‘definite’ macro vascular abnormalities on CTA were identified as true macrovascular causes on IADSA. 11/18 cases identified as ‘possible’ macrovascular causes on CTA were identified as true macrovascular causes on IADSA. In 17 instances, there was discordance between the trained vascular neuroradiology fellow and the clinical report; 11 of these were judged to have a possible macrovascular cause by consensus adjudication with the professor of vascular neuroradiology (HRJ).

In univariable analysis a positive CTA (‘definite’ or ‘possible’ suspicion for a macrovascular cause) has the highest odds ratio for detecting a macrovascular cause (and the greatest area under the curve), followed by: non-deep perforator ICH location; the absence of SVD; the absence of pre-ICH hypertension; and age less than 45 years (Table 2).

Table 2.

Univariable predictors of presence of a macrovascular cause on IADSA.

Predictor variable	OR (95% CI)	ROC AUC (95% CI)
CTA positive^a for vascular malformation	57.5 (21.9–150.5)	0.87 (0.82–0.92)
ICH location		0.66 (0.59–0.74)
deep perforator territory^b	1 (ref)
cerebellar	4.4 (1.5–12.5)
pure intraventricular	6.6 (1.5–29.7)
lobar	4.1 (2.0–8.5)
Absence of confluent SVD on CT	3.2 (1.5–6.7)	0.61 (0.54–0.67)
No pre-ICH hypertension	2.4 (1.2–4.7)	0.59 (0.52–0.65)
Age <45 years	1.3 (0.7–2.4)	0.53 (0.46–0.61)

IADSA: intra-arterial digital subtraction angiography; ICH: intracerebral haemorrhage; CTA: computed tomography angiography; SVD: small vessel disease; OR: odds ratio; ROC: receiver operator curve; AUC: area under the curve.

Positive includes both ‘definite or possible’ suspicion for a macrovascular cause by neuroradiological expert consensus.

Brainstem, basal ganglia, thalamus and caudate nucleus.

We fitted a logistic regression model to the five predictors (with ICH location simplified to non-deep perforator ICH location vs. deep perforator ICH location) to develop a model to predict the risk of a macrovascular cause. Abnormal CTA was the most important predictive factor of the presence of a macrovascular cause in the model (OR 67.4; 95% CI 21.3–213.1, p < 0.001), followed by the absence of SVD (OR 5.0; 95% CI 1.3–19.6, p = 0.019); and the absence of pre-ICH hypertension (OR 3.4; 95% CI 1.0–11.4, p = 0.05). The two weakest predictors – ICH location (OR 2.8; 95% CI 0.96–8.5, p = 0.060) and age <45y (OR 1.1; 95% CI 0.4–3.1, p = 0.90) – were then omitted, which had little impact on the model fit. This simplified model contains just three binary predictors (regression coefficients and model intercept: CTA positive 4.7, No SVD 1.5, no pre-stroke hypertension 1.3, intercept −4.0), and thus can make eight unique predictions, which showed excellent agreement with the actual patient IADSA findings (Hosmer–Lemeshow goodness of fit p = 0.99; no evidence of lack of fit) (Table 3). Bootstrap validation reveals that the ‘optimism-adjusted’¹⁵ ROC area for the simplified model is 0.919 (Figure 2) and that the Cox–Miller calibration¹⁵ slope is 0.949, which suggests that the risk model is well calibrated.

Table 3.

Yield of intracranial macrovascular causes from the optimised model compared to the actual yield.

Predictors			Observed proportion of patients with a macrovascular cause for each combination of predictors in our dataset, % (95% CI) [raw numbers]	Model predicted proportion of patients with a macrovascular cause for each combination of predictors, %
CTA result	SVD on CT	HTN
Positive	No	No	95.3% (84.2–99.4) [41/43]	95.8%
Positive	No	Yes	85.7% (57.2–98.2) [12/14]	86.4%
Positive	Yes	No	80.0% (44.4–97.4) [8/10]	83.7%
Positive	Yes	Yes	75.0% (19.4–99.4) [3/4]	59.0%
Negative	No	No	23.1% (12.5–36.8) [12/52]	22.1%
Negative	No	Yes	5.9% (0.1–28.7) [1/17]	7.3%
Negative	Yes	No	6.3% (0.2–30.2) [1/16]	6.1%
Negative	Yes	Yes	0% [0/17]	1.8%

CTA: computed tomography angiography; HTN: hypertension; SVD: small vessel disease.

‘

Positive’ includes CTA showing a ‘definite’ or ‘possible’ macrovascular cause; ‘negative’ denotes CTA showing no suspicion of a vascular malformation.

Figure 2.

Optimised model ROC curve with abnormal CTA, no pre-stroke hypertension and no moderate to severe SVD on CT used as predictors of macrovascular cause.

The final model allowed us to estimate the risk of a macrovascular cause for any patient based on the combination of CTA result, pre-ICH hypertension and SVD on CT (Table 3). The predicted yield of a macrovascular cause for each combination of predictors is also presented in an algorithm (Figure 3).

Figure 3.

Suggested diagnostic algorithm.

Discussion

In a population of acute ICH patients at risk of an underlying macrovascular cause (having had IADSA based on age, ICH location and vascular risk factors), the presence of small vessel disease (confluent leukoaraiosis and/or one or more lacunes) on non-contrast CT is a useful predictor of the likelihood of finding an underlying macrovascular cause. In our internally validated model, an abnormal CTA, or the combination of a normal CTA with no evidence of SVD (confluent leukoaraiosis on CT) indicate an intermediate to high yield of an underlying macrovascular cause, which might justify undertaking IADSA.

If externally validated on a prospective cohort, the use of our algorithm in clinical practice could lead to more macrovascular causes being detected (higher sensitivity) with an acceptable level of non-diagnostic IADSAs. For instance, if we used our algorithm on our own sample, choosing a predicted yield for a macrovascular cause of >5% as an indication for IADSA, our algorithm would yield 12 more macrovascular causes compared to using CTA alone, at the expense of 85 more negative IADSAs (i.e. an extra 7 patients undertaking an IADSA per additional macrovascular cause identified). In practice, clinicians must balance the expected yield of the IADSA against the procedural risk.

The sensitivity of acutely performed CTA for identifying a macrovascular cause in our study (82%) is lower than reported in a recent meta-analysis of 95% (CI 90–97%),¹⁶ likely due to our use of repeat or delayed IADSA, which revealed a macrovascular cause not seen on the acute IADSA in three instances. Delayed IADSAs were not undertaken in studies involved in the above meta-analysis. Previous studies compared only acute CTA to acute IADSA. However, a recent multicentre prospective study also found a lower sensitivity of acute CTA, in line with our findings.¹⁷ The difference is consistent with an additional yield of macrovascular causes from delayed IADSA, a well-recognised observation in clinical practice, which could be in part related to acute haemorrhage and perihaematomal oedema obscuring small macrovascular lesions. Our study was not able to investigate whether delayed CTA has similar diagnostic accuracy to delayed IADSA, which is a topic for further research.

Although a small case series has previously shown the added diagnostic value of MRI when IADSA is inconclusive,¹⁸ we are only aware of one other study that specifically tests direct visualisation of small vessel disease to improve the prediction of the likelihood of an underlying vascular malformation.¹⁷ In keeping with this study,¹⁷, our data suggest the presence of SVD on acute CT (when CTA is not suspicious for a macrovascular cause) is a useful predictor of a low yield of an underlying macrovascular cause in the investigation of younger patients with spontaneous ICH (mean ages 53 years in the previous study¹⁷ and 50 years in the current study).

Importantly, in our study three patients with an underlying macrovascular cause were older than 45 years with known hypertension and a cerebellar ICH. This combination of variables has previously been associated with a 0% yield of macrovascular causes,⁸ but our data indicate that further investigation looking for a macrovascular cause should be considered in this clinical situation.

Our study has important strengths. All patients identified with a suspected macrovascular cause for ICH at our centre undergo IADSA, so our initial search captured all patients diagnosed with a known macrovascular cause over the study period. Most patients had acute CTA on admission within a standardised care pathway, reducing the potential for selection bias. Our centre has a low threshold for requesting IADSA in suspected macrovascular causes and we were therefore able to include a large cohort of patients with CTA, CT and IADSA (n = 173), which compares favourably in size to previous studies.^12,19–22 The time interval between CTA and the initial DSA was very short, and we included independent blinded CTA ratings by vascular neuroradiologists. Furthermore, we performed delayed and multiple IADSAs in selected patients, which increased the detection of macrovascular causes.

Nevertheless our study also has limitations. The retrospective nature of the study is likely to cause selection and partial verification bias. Although 63% of our cohort had a normal initial CTA (suggesting that there is little partial verification or ‘work up’ bias), our sample is younger than an unselected ICH cohort (median age 48, IQR 40–48), suggesting selection bias, likely due to the fact that IADSA was an inclusion criterion. This reflects current clinical practice with an inherent hesitancy for requesting IADSA in older individuals (e.g. >70 years) in whom that the yield of a macrovascular cause is perceived to be very low.⁸ This selection bias toward younger individuals also explains why our proportion of vascular malformations is much higher than one would expect from an unselected population. Our results can thus only be generalised to younger patients with ICH. This is a difficult bias to overcome in any study, because undertaking IADSA in all unselected consecutive patients with ICH is unlikely to be considered ethical because of the procedural risk, which may be as high as 3% in ICH.⁵ Nevertheless, this acknowledged selection bias is unlikely to weaken our main finding that SVD is predictive of a low yield of a macrovascular abnormality.

We did not capture ICH volume in our cohort, which would have been useful to better characterise our population. In particular the proportion of patients with larger ICH might be relevant, because such patients are less likely to undergo acute IADSA, and can be difficult to classify according to anatomical ICH location.¹⁴ We did not include any patients with multiple acute haemorrhages, presumably because they were considered unlikely to have multiple macrovascular causes and thus did not undergo an IADSA at our centre. Despite these limitations, the demographic features of our study suggest that it is likely to be generalisable to the ICH population that in clinical practice is considered for investigation for an underlying macrovascular cause. A further limitation is that we did not routinely undertake MRI in all included patients, and were therefore unable to investigate additional markers of cerebral SVD underlying ICH, such as cerebral microbleeds, dilated perivascular spaces or cerebral atrophy.^23,24 The use of standardised MRI prior to IADSA would have strengthened our study as it is better able to differentiate between the type of SVD (i.e. cerebral amyloid angiopathy or hypertensive arteriopathy), and is superior for the detection of other underlying causes for ICH (e.g cavernomas or mass lesions). However, the lack of MRI is unlikely to have compromised our ability to detect leukoaraiosis, the main SVD feature analysed in our study: a previous study²⁵ demonstrated that although MRI is more sensitive than CT in detecting subtle (Fazekas grade 1) white matter lesions, MRI and CT have a high concordance in detecting clinically more relevant Fazekas grade 2 and 3 lesions, which we used as predictor in our analysis. Our choice of CT (which was available in all patients acutely at the time of haemorrhage) for assessment of white matter lesions is therefore very unlikely to have led to a significant underestimation of moderate-severe (Fazekas grade ≥2) small vessel disease. Furthermore, our (unpublished) internal inter-rater reliability is moderate to good for the identification of Fazekas 2/3 in ICH populations (κ = 0.60), and we were careful to exclude perihaematomal oedema as a potential mimic of small vessel disease by assessing hemisphere contralateral to the ICH. We did not encounter any cases of transependymal oedema due to acute hydrocephalus, likely due to the selection bias of our patients where we only included those with IADSA, an intervention unlikely to be undertaken in such patients. By contrast, the poor sensitivity and specificity (especially with regards to differentiating enlarged perivascular spaces) of identifying lacunar infarction on CT remains a limitation.

Conclusion

In summary, we show, in a younger acute ICH population (median age 49 years) at risk of having an underlying macrovascular cause, the identification of SVD (moderately severe leukoaraiosis or lacunar infarction on CT), in combination with CTA and pre-ICH hypertension ICH, can predict which patients have a low yield of an intracranial macrovascular cause. If externally validated, our algorithm has potential to more effectively target invasive IADSA towards patients with the highest potential diagnostic yield, avoiding unnecessary invasive tests and improving the diagnostic pathway for patients with spontaneous ICH.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: DW and DJW received finding from the stroke association and the British heart foundation. GA received funding from the National Institute for Health Research University College London Hospitals Biomedical Research Centre.

Ethical approval

Institutional Review Board approval was obtained.

Informed consent

Written informed consent was waived by the Institutional Review Board.

Guarantor

DJW.

Contributorship

DJW and HRJ contributed equally to this work. DW, DJW, IJ and HRJ were involved with study concept, DW, DJW and HRJ were involved with data retrieval and data interpretation, DW and GA were involved with statistical analysis. AO was involved with data interpretation. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

References

Poon

Fonville

Al-Shahi Salman

. Long-term prognosis after intracerebral haemorrhage: Systematic review and meta-analysis. J Neurol Neurosurg Psychiatry 2014; 85: 660–667.

van Asch

Luitse

Rinkel

et al.

Incidence, case fatality, and functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: A systematic review and meta-analysis. Lancet Neurol 2010; 9: 167–176.

Foulkes

Wolf

Price

et al.

The Stroke Data Bank: Design, methods, and baseline characteristics. Stroke 1988; 19: 547–554.

Pantoni

. Cerebral small vessel disease: From pathogenesis and clinical characteristics to therapeutic challenges. Lancet Neurol 2010; 9: 689–701.

Dawkins

Evans

Wattam

et al.

Complications of cerebral angiography: a prospective analysis of 2,924 consecutive procedures. Neuroradiology 2007; 49: 753–759.

Cordonnier

Klijn

van Beijnum

et al.

Radiological investigation of spontaneous intracerebral hemorrhage: Systematic review and trinational survey. Stroke 2010; 41: 685–690.

Toffol

Biller

Adams

Jr et al.

The predicted value of arteriography in nontraumatic intracerebral hemorrhage. Stroke 1986; 17: 881–883.

Zhu

Chan

Poon

. Spontaneous intracranial hemorrhage: Which patients need diagnostic cerebral angiography? A prospective study of 206 cases and review of the literature. Stroke 1997; 28: 1406–1409.

Halpin

Britton

Byrne

et al.

Prospective evaluation of cerebral angiography and computed tomography in cerebral haematoma. J Neurol Neurosurg Psychiatry 1994; 57: 1180–1186.

10.

Delgado Almandoz

Jagadeesan

Moran

et al.

Independent validation of the secondary intracerebral hemorrhage score with catheter angiography and findings of emergent hematoma evacuation. Neurosurgery 2012; 70: 131–140.

11.

van Asch

Velthuis

Greving

et al.

External validation of the secondary intracerebral hemorrhage score in The Netherlands. Stroke 2013; 44: 2904–2906.

12.

Delgado Almandoz

Schaefer

Forero

et al.

Diagnostic accuracy and yield of multidetector CT angiography in the evaluation of spontaneous intraparenchymal cerebral hemorrhage. AJNR Am J Neuroradiol 2009; 30: 1213–1221.

13.

Wahlund

Barkhof

Fazekas

et al.

A new rating scale for age-related white matter changes applicable to MRI and CT. Stroke 2001; 32: 1318–1322.

14.

Charidimou A, Schmitt A, Wilson D, et al. The Cerebral Haemorrhage Anatomical RaTing inStrument (CHARTS): Development and assessment of reliability. J Neurol Sci 2017; 372: 178–183.

15.

Steyerberg

. Clinical prediction models, New York: Springer, 2008.

16.

Josephson

White

Krishan

et al.

Computed tomography angiography or magnetic resonance angiography for detection of intracranial vascular malformations in patients with intracerebral haemorrhage. Cochrane Database Syst Rev 2014; 9: CD009372.

17.

van Asch

Velthuis

Rinkel

et al.

Diagnostic yield and accuracy of CT angiography, MR angiography, and digital subtraction angiography for detection of macrovascular causes of intracerebral haemorrhage: Prospective, multicentre cohort study. BMJ 2015; 351: h5762.

18.

Lummel

Lutz

Bruckmann

et al.

The value of magnetic resonance imaging for the detection of the bleeding source in non-traumatic intracerebral haemorrhages: A comparison with conventional digital subtraction angiography. Neuroradiology 2012; 54: 673–680.

19.

Romero

Artunduaga

Forero

et al.

Accuracy of CT angiography for the diagnosis of vascular abnormalities causing intraparenchymal hemorrhage in young patients. Emerg Radiol 2009; 16: 195–201.

20.

Wong

Siu

Abrigo

et al.

Computed tomographic angiography and venography for young or nonhypertensive patients with acute spontaneous intracerebral hemorrhage. Stroke 2011; 42: 211–213.

21.

Yoon

Chang

Choi

et al.

Multidetector row CT angiography in spontaneous lobar intracerebral hemorrhage: A prospective comparison with conventional angiography. AJNR Am J Neuroradiol 2009; 30: 962–967.

22.

Yeung

Ahmad

Aviv

et al.

Comparison of CTA to DSA in determining the etiology of spontaneous ICH. Can J Neurol Sci 2009; 36: 176–180.

23.

Fazekas

Kleinert

Roob

et al.

Histopathologic analysis of foci of signal loss on gradient-echo T2*-weighted MR images in patients with spontaneous intracerebral hemorrhage: Evidence of microangiopathy-related microbleeds. AJNR Am J Neuroradiol 1999; 20: 637–642.

24.

Shoamanesh

Kwok

Benavente

. Cerebral microbleeds: Histopathological correlation of neuroimaging. Cerebrovasc Dis 2011; 32: 528–534.

25.

Wattjes

Henneman

van der Flier

et al.

Diagnostic imaging of patients in a memory clinic: Comparison of MR imaging and 64-detector row CT. Radiology 2009; 253: 174–183.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB

Developing an algorithm to identify patients with intracerebral haemorrhage secondary to a macrovascular cause

Abstract

Introduction

Patients and methods

Results

Discussion and conclusion

Keywords

Introduction

Patient and methods

Patient selection

Data collection

Image analysis

Statistical analysis

Ethical approval

Results

Discussion

Conclusion

Footnotes

Declaration of Conflicting Interests

Funding

Ethical approval

Informed consent

Guarantor

Contributorship

References

Supplementary Material