Old Microbleeds Are a Potential Risk Factor for Cerebral Bleeding After Ischemic Stroke
A Gradient-Echo T2*-Weighted Brain MRI Study
Background and Purpose— T2*-weighted gradient-echo MRI is known to detect old microbleeds (MBs), considered indicative of microangiopathy. MBs might be a potential risk factor for early cerebral bleeding (CB) after ischemic stroke. Therefore, we assessed the impact of MBs on the occurrence of CB after cerebral infarction.
Methods— We included prospectively stroke patients who had documented ischemic damage. The imaging protocol involved baseline CT scan, T2*-weighted gradient-echo MRI, diffusion-weighted imaging, T2-weighted imaging, and magnetic resonance angiography and had to be performed within 24 hours after symptom onset. The assessment of CB with T2*-weighted gradient-echo sequence necessitated a focal area of signal loss either within the ischemic area revealed by diffusion-weighted imaging or remote from it. Old MBs were defined on T2*-weighted images as homogeneous rounded areas of signal loss without surrounding edema. CT scan was systematically repeated within the first week to verify CB as diagnosed by the T2* weighted sequence.
Results— One hundred patients (mean age, 60±13 years; range, 19 to 83 years; 58 men, 42 women) met the inclusion criteria. MBs were seen in 20 patients on T2*-weighted imaging. Multivariate logistic regression analysis revealed that age, diabetes, previous use of antithrombotic drugs, evidence of an atherothrombotic source of stroke, and lacunar infarct were significantly associated with MBs (P<0.0001). CB was diagnosed in 26 patients: at the acute stage by T2*-gradient echo sequence in 18 patients and with CT scan performed within the first week in 8 patients. Multivariate logistic regression analysis showed that baseline National Institutes of Health Stroke Scale score, diabetes, and MBs were considered significant and independent predictors of CB (P<0.001).
Conclusions— Although the pathogenesis of CB after ischemic stroke is multifactorial, the increased observation of CB in patients with MBs suggests that the associated vascular vulnerability contributes to CB.
Approximately 20% to 40% of all stroke patients experience hemorrhagic transformation within the first week after symptom onset.1 Although cerebral bleeding (CB) is a common event that occurs independently of therapy,2 caution is required when thrombolysis3,4⇓ or anticoagulants5 are administered.
The identification of risk factors for CB might be helpful in improving the risk-to-benefit ratio of thrombolytic treatments.6–9⇓⇓⇓ Hypertension, embolic origin, use of anticoagulant therapy, increasing stroke severity, and duration and intensity of the cerebral blood flow reduction have been associated with a higher risk of CB.10–13⇓⇓⇓
New MRI techniques may also identify patients at increased risk of subsequent intracranial hemorrhage.14 A higher sensitivity of MRI to hyperacute hemorrhage15–18⇓⇓⇓ lasting <12 hours has been demonstrated when T2*-weighted sequences are used. Moreover, T2*-weighted sequences have the potential to reveal old intracranial bleeding in a substantial proportion of patients with primary lobar hemorrhage19,20⇓ and less frequently in patients with ischemic stroke.21 These microbleeds (MBs) are thought to be indicative of microangiopathy. MBs are also considered as a marker of amyloid angiopathy.22
The likelihood of early CB after ischemic stroke might be increased in patients who had the most vulnerable microvascular system.23 MRI demonstration of MBs could gain even more clinical significance if this finding could be used to identify patients at increased risk of early CB. Therefore, we assessed the impact of this abnormality on the occurrence of CB.
Patients and Methods
We included prospectively ischemic stroke patients whose symptoms lasted >1 hour and who had documented ischemic damage. Stroke onset was defined as the last time the patient was known to be free of deficit. All patients were examined acutely by a stroke attending physician. The NIH Stroke Scale (NIHSS) score24 was recorded by a clinician certified in the administration of the scale at the time of baseline CT imaging and was repeated on day 7.
The multisequence MRI protocol included at least diffusion-weighted imaging (DWI), T2-weighted imaging, T2*-weighted gradient echo, and magnetic resonance angiography and had to be performed as a single session within 24 hours after symptom onset. Patients with unstable vital signs or general MRI contraindications were excluded from the MRI study. Informed consent was required from all patients or their next of kin. Patients were included in the analysis if they had a readable MRI.
Antithrombotic drugs were given according to the suspected mechanism of stroke and current therapeutical management in use at our institution. When tissue plasminogen activator (tPA) was given, MRI was performed after thrombolysis to avoid excessive delay to treatment. Other antithrombotic drugs were administered after either the CT scan or MRI if the latter was obtained immediately after the CT scan. Acute neurological worsening led to an immediate CT scan. CB was considered spontaneous if the patient did not receive any antithrombotic drugs before MRI. CBs were categorized into 4 different subtypes according to the Second European Cooperative Acute Stroke Study (ECASSII) criteria.8 In addition to the pure radiological definition, the category of symptomatic hemorrhage was used for patients with clinical deterioration of ≥4 points on the NIHSS and with no CT findings that might have been responsible for this deterioration other than a hemorrhage. Neurovascular workup included analysis of risk factors, including history of hypertension, diabetes mellitus, hyperlipidemia, smoking habit, stroke or TIA, cardiac disease, previous use of antithrombotic agents, and search for an arterial or cardiac source of embolism with neck ultrasound, ECG, and echocardiography. The Trial of Org 10172 in Acute Stroke treatment (TOAST) criteria were used to classify stroke type.25
Imaging Data Acquisition
MRI was performed with a superconductive unit operating at 1.5 T (Siemens AG, Medical Group) that used a circular polarized head coil equipped with enhanced gradient hardware for echo-planar imaging. The MRI procedure included 3 sequences. First was a dual fast spin-echo T2 sequence with an echo time (TE) of 16 to 98 ms, repetition time (TR) of 3000 ms, 20 axial slices, thickness of 5 mm, distance factor of 0.20, asymmetric matrix of n×512, field of view of 250 mm, and acquisition time of 5 minutes. Second was an axial isotropic DWI spin-echo echo-planar imaging sequence with a TR of 4.700 ms, TE of 118 ms, 20 axial slices, thickness of 5 mm, matrix of 128×128, field of view of 260 mm, 1 excitation, and an acquisition time of 23 seconds. Two b values were used (0 and 1000 s/mm2) in the vision system. The DWI sequence at b=1.000 was run 3 times with diffusion gradients applied in each of the x, y, and z directions. The third sequence was a T2*-weighted gradient-echo sequence with a TR of 800 ms, TE of 26 ms, flip angle of 20°, thickness of 5 mm, 20 axial slices, distance factor of 0.20, asymmetric matrix of n×256, 2 excitations, acquisition time of 6 minutes, and a field of view of 250 mm. The final sequence was a 3DTOF turbo magnetic resonance angiography with a TR of 35 ms, TE of 6.4 ms, flip angle of 20°, asymmetric matrix of n×512, field of view of 240 mm, 1 excitation, and an acquisition time of 6 minutes 44 seconds, with 3 axial slabs (thickness, 31.9 mm; partitions, 24; distance factor, 0.38) placed over the entire circle of Willis. The scanning time for the whole MRI protocol was ≈20 minutes with an additional 10 minutes for patient positioning.
The analysis involved 4 stages. In the first step, assessment of CB with T2*-weighted gradient-echo sequences necessitated a focal area of signal loss consistent with bleeding either within the ischemic area revealed by DWI or remote from the ischemic damage (Figure 1). Second, MBs were defined on T2*-weighted images as homogeneous rounded areas of signal loss of 2 to 5 mm in diameter without surrounding edema (Figure 2). An area of symmetric hypointensity of the globus pallidus, likely to represent calcification, and loss of signal in the distal middle cerebral artery (MCA) and its branches seen in the sylvian fissure consistent with calcified atheromatous plaque or acute thrombosis (MCA dot sign)26 were disregarded. Third, the severity of a white matter hyperintense area on T2-weighted images (Figure 2) was graded according to the method of Fazekas and colleagues.27 Finally, lacunes were categorized as areas (<10 mm in diameter) isointense to cerebrospinal fluid and hyperintense on T2-weighted sequences (Figure 2).28
Baseline CT scan was performed with a fourth-generation CT (Elscint CT Twin). Thrombolytic therapy monitoring in our protocol included a follow-up CT scan at days 1 and 7 as previously described.29 A repeated CT scan within the first week was required for patients who received other antithrombotic drugs. CT scans and MRI were reviewed by 2 neuroradiologists without knowledge of clinical data and treatment assignment (M.H., J.C.F.) whose consensus determined the MRI and CT findings.
Logistic regression analysis was used to assess (1) the relationship between MBs with the variables age, sex, hypertension, diabetes, hyperlipidemia, smoking, previous stroke, cardioembolic and atherothrombotic source of stroke, lacunae, white matter hyperintensity, and long-term use of antithrombotics and (2) the relationship between CB and the variables age, sex, baseline NIHSS score, hypertension, diabetes, hyperlipidemia, previous stroke, long-term use of antithrombotics, cardioembolic and atherothrombotic source of stroke, lacunae, MBs, white matter hyperintensity, and treatment. For MBs and CB, we have fitted univariate logistic regression model for each explanatory variable considered, with estimated OR, 95% CI of the theoretical OR, and the probability value of the likelihood ratio test. To select 1 multivariate logistic regression model, we first fitted the full model, including all explanatory variables, and then we used an interactive backward elimination method based on the likelihood ratio test. The probability value used for either excluding or entering variables was fixed at 0.10. Mean values were expressed as mean±SD. A subgroup analysis comparing patients submitted to MRI before versus after treatment (thrombolytic or antithrombotic therapy) was performed using 2 proportion comparison tests with Fisher’s exact test. This analysis was performed both for CB and MBs. All data were analyzed with the Statistical Package of Social Science (version 10) software.
Between January 2000 and May 2001, 100 patients (mean age, 60±13 years; range, 19 to 83 years; 58 men, 42 women) met the inclusion criteria. During this same period, 810 patients were evaluated by the inpatient Lyon Stroke Service, and 572 subsequently received the diagnosis of acute ischemic stroke. Among the 100 patients, 53% had hypertension, 33% had hypercholesterolemia, 19% had diabetes, 47% were smokers, 26% had a history of transient ischemic attack or minor stroke, and 27% were long-term users of antithrombotic drugs.
Stroke involved anterior circulation in 60 patients and posterior circulation in 40 patients. Mean baseline NIHSS score was 10±6.3. CT scan was performed 4±4 hours (median, 3 hours) and MRI was done 10±7 hours (median, 9 hours) after stroke onset. Twenty-seven patients received tPA before MRI assessment, whereas 73 patients were given other antithrombotic agents—58 were treated before MRI and 15 after MRI. Eight patients received antiplatelet agents (aspirin, thienoperidin), 39 received efficient anticoagulant therapy (heparin infusion or warfarin), and 26 had subcutaneous low-molecular-weight heparin (Nadroparin 0.3 mL, once or twice daily). Stroke was attributed to large-artery atherosclerosis in 37 patients, cardioembolic source in 30, small-vessel disease in 15, arterial dissection in 4, and unknown origin in 14.
MBs were seen in 20 patients exclusively on T2*-weighted imaging. MBs were found in 9 patients who had MRI before therapy and in 11 patients who had MRI after treatment initiation. No significant difference was recorded between groups (P=0.80; CI, −0.14 to 0.18). Most patients with MBs exhibited multiple lesions, which were located mainly in cortical-subcortical regions and basal ganglia. Univariate logistic regression analysis (Table 1) showed that patients with MBs were significantly older, hypertensive, and diabetic and had significantly more lacunar infarcts and atherothrombotic source of stroke. MBs were also more frequent in patients who were long-term antithrombotic users. Multivariate logistic regression analysis (Table 2) revealed that age, diabetes, previous use of antithrombotic drugs, evidence of an atherothrombotic source of stroke, and lacunar infarct were significantly associated with MBs.
CB was diagnosed in 26 patients at the acute stage by T2*-weighted gradient-echo sequence in 18 patients and with CT scan performed within the first week in 8 patients. Suspected bleeding on T2*-weighted gradient-echo sequence was confirmed by follow-up CT. Bleeding was observed within the ischemic area (n=24) and remote from it (n=2). CB was defined as hemorrhagic infarction (n=21; type 1, n=13; type 2, n=8) and parenchymal hemorrhage (n=5; type 1, n=2; type 2, n=3). Clinical deterioration was related to parenchymal hemorrhage type 2 (n=3) defined as dense hematoma of >30% of the infarcted area. Death occurred in 6 patients and was caused by either the progression of brain ischemia (n=5) or recurrent stroke (n =1). Univariate logistic regression analysis (Table 3) showed that patients who experienced CB were older, more often were hypertensive and diabetic, and more frequently had lacunes and MBs (Old MBs were detected in 10 patients who had CB). No significant difference in CB rate was observed between the different therapeutic groups, even after adjustment for all variables (P=0.7). CB occurred in 16 patients who had therapy before MRI and in 10 patients who were treated after MRI. A significant difference in bleeding rate was recorded between groups (P=0.002; CI, 0.03 to 0.38), thus suggesting a possible role of antithrombotic drugs. Multivariate logistic regression analysis (Table 4) showed that baseline NIHSS score, diabetes, and more clearly MBs were considered significant and independent predictors of CB.
CB can be a devastating complication of ischemic stroke and the main adverse effect of thrombolytic treatment.7,8,30,31⇓⇓⇓ Conventional MRI often fails to detect CB at the early stage of stroke. As a result, CT scan remains the standard diagnostic test for identification of CB during this period. New MRI techniques provide critical information in detecting acute bleeding.15–18⇓⇓⇓ As blood extravasates in the tissue, the hemoglobin molecule becomes deoxygenated. Deoxyhemoglobin thereby produces a nonuniform magnetic field that results in rapid dephasing of proton spins in T2- and more so in T2*-weighted images. To the best of our knowledge, this study is the earliest evaluation of acute CB with T2*-weighted imaging in a large cohort of ischemic stroke patients. Our results confirm the usefulness of T2*-weighted gradient-echo sequence in detecting early hemorrhage transformation as part of a multimodal stroke MRI protocol.
The mechanism of CB remains a remarkably complex and dynamic process involving a combination of microvascular injury with altered permeability and reperfusion integrated over time.23,32,33⇓⇓ The spectrum of CB differs widely and may include some trivial hemorrhagic petechiae or parenchymal hemorrhage with space-occupying effect with an increased risk of clinical worsening. As previously described,34,35⇓ clinical deterioration was related to parenchymal hemorrhage type 2 (n=3) defined as dense hematoma of >30% of the infarcted area
Brain microvasculature is potentially weakened by such factors as increasing age, sustained exposure to elevated blood pressure, hyperglycemia, and amyloid or fibrohyalinosis degeneration of brain blood vessels. Old MBs provide further evidence of severe microangiopathy with a subsequent increased vascular vulnerability.
Roob et al36 used gradient-echo MRI to detect small hemosiderin deposits in 280 people without clinical neurological disease. MBs were found in 6% of these individuals and were associated with advancing age, hypertension, and leukoariosis. Histological examination of MR foci of signal loss has detected moderate to severe fibrohyalinosis, suggesting that lacunar hemorrhages are related to bleeding-prone microangiopathy.37–39⇓⇓
The rate of hemosiderin deposits was in the range of previous study using T2*-weighted gradient-echo sequence21 In line with current data, we found a significant association between the presence of MBs and other morphological signs of cerebral microangiopathy. The association of long-term use of antithrombotic agents was also significantly linked to the presence of MBs
The risk of intracerebral hemorrhage after secondary prevention with oral anticoagulants seems higher in patients anticoagulated after a nondisabling ischemic stroke than after myocardial infarction.40 Leukoariosis and lacunar state of the basal ganglia are frequently observed in patients who experienced intracranial hemorrhage under warfarin therapy. Both abnormalities have been suggested to indicate a higher risk of bleeding.41 However, in these CT scan studies, the rate of MBs underlying the lacunar aspect could not be assessed. According to our data, old MBs might also be involved in the group of known risk factors favoring CB after brain ischemia. However, several limitations must be considered. The rate of hemorrhage transformation might have been underestimated because the infarcts could have become hemorrhagic later and thus were not seen on follow-up CT scan. Although most secondary hemorrhages occur within the second week after ischemic stroke,42 unfavorable outcomes are usually expected within the early period, when the assessment of bleeding by T2* MRI might influence stroke management. Other possible causes of areas of hypointensity such as a cavernous hemangioma may have been assessed as old lacunar hemorrhage. Moreover, because of the small sample size of each therapeutic group, our study did not have the statistical power to demonstrate the hemorrhagic effect of a particular treatment. Conversely, a global hemorrhagic effect can be anticipated from the subgroup analysis of patients who had MRI after treatment.
Owing to the saturation of our MR unit (a single machine for an 865-bed third-referral center), only 100 patients of 572 stroke subjects could be included at the acute stage. Moreover, a total of 27% received recombinant tPA; this high frequency was related to our recruitment procedure that focused on early admission consistent with the use of recombinant tPA. Accordingly, it is conceivable that this sample might be not representative of the whole stroke population, thus hampering interpretation of these data Although the pathogenesis of CB is multifactorial, the relationship between acute CB and previous vessel wall damage as revealed by the presence of MBs suggests an increased vascular vulnerability. However, larger numbers of patients in general and those with MBs in particular are needed to clearly established this relation. Although the results of the analysis comparing the rate of MBs between patients who had MRI before and after treatment were not significant, the distinction between hemosiderin deposits and more recent bleedings needs a preposttreatment T2*-weighted MRI assessment. However, because the benefit of antithrombotic agents outweighs the hemorrhagic risk in randomized clinical trials, further research is needed to examine the risk-to-benefit ratio of antithrombotics in a subgroup at risk of both thrombotic and hemorrhagic event. For this purpose, the use of MRI as baseline imaging modal ity is warranted.
We are grateful to the HCL Research Office and Siemens Society for supporting this work, to Health Library Office staff members at UCB Lyon 1 (Monique Billaud) and IFNL (Chantal Beranger) for the figures, and to Michèle Canova for editing assistance. This study was funded by ARNEI.
- Received September 27, 2001.
- Revision received November 22, 2001.
- Accepted November 30, 2001.
- ↵Moulin T, Crepin Leblond T, Chopard JL, Bogousslavsky J. Hemorrhagic infarcts. Eur Neurol. 1993; 34: 64–77.
- ↵Toni D, Fiorelli M, Bastianello S, Sacchetti ML, Sette G, Argentino C, Montinaro E, Bozzao L. Hemorrhagic transformation of brain infarct: predictability in the first five hours from stroke onset and influence on clinical outcome. Neurology. 1996; 46: 341–345.
- ↵Slivka A, Pulsinelli WA. Hemorrhagic complications of thrombolytic therapy in experimental stroke. Stroke. 1987; 18: 1146–1148.
- ↵Babikian VL, Kase CS, Pessin MS, Norrving B, Gorelick PB. Intracerebral hemorrhage in stroke patients anticoagulated with heparin. Stroke. 1989; 20: 1500–1503.
- ↵Motto C, Ciccone A, Aritzu E, Boccardi E, De Grandi C, Piana A, Candelise L, for the MAST-I Collaborative Group. Hemorrhage after an acute ischemic stroke. Stroke. 1999; 30: 761–764.
- ↵NINDS t-PA Stroke Study Group. Intracerebral hemorrhage after intravenous t-PA therapy for ischemic stroke. Stroke. 1997; 28: 2109–2118.
- ↵Larrue V, von Kummer R, Del Zoppo G, Bluhmki E. Hemorrhagic transformation in acute ischemic stroke: potential contributing factors in the European Cooperative Acute Stroke Study. Stroke. 1997; 205: 327–333.
- ↵Tong DC, Adami A, Moseley ME, Marks MP. Relationship between apparent diffusion coefficient and subsequent hemorrhagic transformation following acute ischemic stroke. Stroke. 2000; 31: 2378–2384.
- ↵Alexandrov AV, Black SE, Ehrlich LE, Cladwell CB, Norris JW. Predictors of hemorrhagic transformation occurring spontaneously and on anticoagulants in patients with acute ischemic stroke. Stroke. 1997; 28: 1198–1202.
- ↵Tong DC, Adami A, Moseley ME, Marks MP. Relationship between apparent diffusion coefficient and subsequent hemorrhagic transformation following acute ischemic stroke. Stroke. 2000; 31: 2378–2384.
- ↵Patel MR, Edelman RR, Warach S. Detection of hyperacute primary intraparenchymal hemorrhage by magnetic resonance imaging. Stroke. 1996; 27: 2221–2324.
- ↵Schellinger PD, Jansen O, Fiebach JB, Hacke W, Sartor K. A standardized MRI protocol: comparison with CT in hyperacute intracerebral hemorrhage. Stroke. 1999; 30: 765–768.
- ↵Linfante I, Llinas RH, Caplan LR, Warach S. MRI features of intracerebral hemorrhage within 2 hours from symptom onset. Stroke. 1999; 30: 2263–2267.
- ↵Greenberg S, Finkelstein S, Schaefer P. Petechial hemorrhages accompanying lobar hemorrhage: detection by gradient echo MRI. Neurology. 1996; 46: 1751–1754.
- ↵Fazekas F, Kleinert R, Roob G, Kleinert G, Kapeller P, Schmidt R. Histopathological 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.
- ↵Greenberg S, O’Donnell H, Schaefer P, Kraft E. MRI detection of new hemorrhages: potential marker of progression in cerebral amyloid angiopathy. Neurology. 1999; 22: 1135–1138.
- ↵Del. Zoppo G, Von Kummer R, Hamman GF. Ischaemic damage of brain microvessels: inherent risks for thrombolytic treatment in stroke. J Neurol Neurosurg Psychiatry. 1998; 65: 1–9.
- ↵Brott TG, Adams HP, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, Holleran R, Eberle R, Hertzberg V, Rorick M, Moomav CJ, Walker M. Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989; 20: 864–870.
- ↵Adams HP Jr, Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon DL, Marsh EEIII. Classification of subtype of acute ischemic stroke: definitions for use in a multicenter clinical trial: TOAST: Trial of Org 10172 in Acute Stroke treatment. Stroke. 1993; 24: 35–41.
- ↵Barber PA, Demchuk AM, Hudon ME, Warwick Pexman JH, Hill MD, Buchan A M. Hyperdense sylvian fissure MCA “dot” sign: a CT marker of acute ischemia. Stroke. 2001; 32: 84–88.
- ↵Fazekas F, Chawluk JB, Alavi A, Hurtig HI, Zimmerman RA. MR signal abnormalities at 1.5T in Alzheimer’s dementia and normal aging. AJNR Am J Neuroradiol. 1987; 8: 421–426.
- ↵Offenbacher H, Fazekas F, Schmidt R, Koch M, Fazekas G, Kapeller P. MR of cerebral abnormalities concomitant with primary intracerebral hematomas. AJNR Am J Neuroradiol. 1996; 17: 573–578.
- ↵Trouillas P, Nighoghossian N, Derex L, Adeleine P, Honnorat J, Neuschwander P, Riche G, Getenet JC, Wei Li, Froment JC, Turjman F, Malicier D, Fournier G, Gabry AL, Ledoux X, Berthezène Y, French P, Dechavanne M. Thrombolysis with intravenous rtPA in a series of 100 cases of acute carotid artery stroke: determination of etiological, topographic, and radiological outcome factors. Stroke. 1998; 29: 2529–2540.
- ↵Jaillard A, Cornu C, Durieux A, Moulin T, Bouititie F, Lees KR, Hommel M, for the MAST-E Group. Hemorrhagic transformation in acute ischemic stroke: the MAST-E Study. Stroke. 1999; 30: 1326–1332.
- ↵Larrue V, Von Kummer R, Müller A, Bluhmki E. Risk Factors for severe hemorrhagic transformation in ischemic stroke patients treated with recombinant tissue plasminogen activator: a secondary analysis of the European-Australasian Acute Stroke Study (ECASSII). Stroke. 2001; 32: 438–441.
- ↵Hart RG, Easton JD. Hemorrhagic infarcts. Stroke. 1986; 17: 586–589.
- ↵Fiorelli M, Bastianello S, von Kummer R, del Zoppo GJ, Larrue V, Lesaffre E, Ringleb AP, Lorenzano S, Manelfe C, Bozzao L. Hemorrhagic transformation within 36 hours of a cerebral infarct: relationships with early clinical deterioration and 3-month outcome in the European Cooperative Acute Stroke Study I (ECASSI) cohort. Stroke. 1999; 30: 2280–2284.
- ↵Berger C, Fiorelli M, Steiner T, Schä bitz, Bozzao L, Bluhmki, Hacke W, von Kummer R. Hemorrhagic transformation of ischemic brain tissue: asymptomatic or asymptomatic? Stroke. 2001; 32: 1330–1335.
- ↵Roob G, Schmidt R, Kapeller P, Lechner A, Hartung HP, Fazekas F. MRI evidence of past microbleeds in a healthy elderly population. Neurology. 1999; 52: 991–994.
- ↵Fazekas F, Kleinert R, Offenbacher H, Schmidt R, Kleinert G, Payer F, Radner H, Lechner H. Pathologic correlates of incidental MRI white matter signal hyperintensities. Neurology. 1993; 43: 1683–1689.
- ↵Roob G, Lechner A, Schmidt R, Flooh E, Hartung HP, Fazekas F. Frequency and location of microbleeds in patients with primary intracerebral hemorrhage. Stroke. 2000; 31: 2665–2669.
- ↵Kinoshita T, Okudera T, Tamura H, Ogawa T, Hatazawa J. Assessment of lacunar hemorrhage associated with hypertensive stroke by echo-planar gradient-echo T2*-weighted MRI. Stroke. 2000; 31: 1646–1650.