Ensemble of Convolutional Neural Networks Improves Automated Segmentation of Acute Ischemic Lesions Using Multiparametric Diffusion-Weighted MRI.

BACKGROUND AND PURPOSE: Accurate automated infarct segmentation is needed for acute ischemic stroke studies relying on infarct volumes as an imaging phenotype or biomarker that require large numbers of subjects. This study investigated whether an ensemble of convolutional neural networks trained on multiparametric DWI maps outperforms single networks trained on solo DWI parametric maps. MATERIALS AND METHODS: Convolutional neural networks were trained on combinations of DWI, ADC, and low b-value-weighted images from 116 subjects. The performances of the networks (measured by the Dice score, sensitivity, and precision) were compared with one another and with ensembles of 5 networks. To assess the generalizability of the approach, we applied the best-performing model to an independent Evaluation Cohort of 151 subjects. Agreement between manual and automated segmentations for identifying patients with large lesion volumes was calculated across multiple thresholds (21, 31, 51, and 70 cm3). RESULTS: An ensemble of convolutional neural networks trained on DWI, ADC, and low b-value-weighted images produced the most accurate acute infarct segmentation over individual networks (P < .001). Automated volumes correlated with manually measured volumes (Spearman ρ = 0.91, P < .001) for the independent cohort. For the task of identifying patients with large lesion volumes, agreement between manual outlines and automated outlines was high (Cohen κ, 0.86-0.90; P < .001). CONCLUSIONS: Acute infarcts are more accurately segmented using ensembles of convolutional neural networks trained with multiparametric maps than by using a single model trained with a solo map. Automated lesion segmentation has high agreement with manual techniques for identifying patients with large lesion volumes.

The Causative Classification of Stroke system: an international reliability and optimization study.

BACKGROUND: Valid and reliable ischemic stroke subtype determination is crucial for well-powered multicenter studies. The Causative Classification of Stroke System (CCS, available at http://ccs.mgh.harvard.edu) is a computerized, evidence-based algorithm that provides both causative and phenotypic stroke subtypes in a rule-based manner. We determined whether CCS demonstrates high interrater reliability in order to be useful for international multicenter studies. METHODS: Twenty members of the International Stroke Genetics Consortium from 13 centers in 8 countries, who were not involved in the design and development of the CCS, independently assessed the same 50 consecutive patients with acute ischemic stroke through reviews of abstracted case summaries. Agreement among ratings was measured by kappa statistic. RESULTS: The κ value for causative classification was 0.80 (95% confidence interval [CI] 0.78-0.81) for the 5-subtype, 0.79 (95% CI 0.77-0.80) for the 8-subtype, and 0.70 (95% CI 0.69-0.71) for the 16-subtype CCS. Correction of a software-related factor that generated ambiguity improved agreement: κ = 0.81 (95% CI 0.79-0.82) for the 5-subtype, 0.79 (95% CI 0.77-0.80) for the 8-subtype, and 0.79 (95% CI 0.78-0.80) for the 16-subtype CCS. The κ value for phenotypic classification was 0.79 (95% CI 0.77-0.82) for supra-aortic large artery atherosclerosis, 0.95 (95% CI 0.93-0.98) for cardioembolism, 0.88 (95% CI 0.85-0.91) for small artery occlusion, and 0.79 (0.76-0.82) for other uncommon causes. CONCLUSIONS: CCS allows classification of stroke subtypes by multiple investigators with high reliability, supporting its potential for improving stroke classification in multicenter studies and ensuring accurate means of communication among different researchers, institutions, and eras.

Reversible cerebral vasoconstriction syndrome with posterior leucoencephalopathy after oral contraceptive pills.

Reversible cerebral vasoconstriction syndrome (RCVS) is characterized by sudden-onset recurrent 'thunderclap' headaches with reversible multifocal narrowing of the cerebral arteries, often associated with focal neurological deficits from ischaemic or haemorrhagic stroke. It has been associated with exposure to vasoconstrictive drugs, pregnancy, migraine, and a variety of other conditions. Whereas the pathophysiology of RCVS remains unclear, changes in the levels of female hormones are considered important because RCVS predominantly affects women and is frequently associated with pregnancy. We report a patient with angiographically confirmed RCVS whose MRI showed reversible brain oedema, suggesting an overlap between RCVS and the reversible posterior leucoencephalopathy syndrome. The only identified risk factor was oral contraceptive pills started 1 month prior to onset, supporting a role for female reproductive hormones in precipitating this overlap syndrome.

Functional contrast-enhanced CT for evaluation of acute ischemic stroke does not increase the risk of contrast-induced nephropathy.

BACKGROUND AND PURPOSE: Concerns have recently grown regarding the safety of iodinated contrast agents used for CTA and CTP imaging. We tested whether the incidence of AN, defined by a >or=25% increase in the post-contrast scan creatinine level, was higher among patients with ischemic stroke who underwent a functional contrast-enhanced CT protocol compared with those who had no iodinated contrast administration. MATERIALS AND METHODS: The contrast-exposed group consisted of 575 patients with acute ischemic stroke who underwent CTA (n = 313), CTA/CTP (n = 224), or CTA/CTP followed by conventional angiography (n = 38) within 24 hours of stroke onset and were consecutively enrolled in a prospective cohort study. The nonexposed group consisted of 343 patients with ischemic stroke, consecutively admitted to the same institution, who did not receive iodinated contrast material. Patients were stratified by baseline eGFR. In the primary analysis, the Fisher exact test was used to compare the incidence of AN between the contrast-exposed and the nonexposed patients at 24, 48, and 72 hours and on a cumulative basis. A secondary analysis compared the incidence of AN in patients who underwent conventional angiography following CTA/CTP versus patients who underwent CTA/CTP only. RESULTS: The incidence of AN was 5% in the exposed and 10% in the nonexposed group (P = .003). Patients who underwent conventional angiography after contrast CT were at no greater risk of AN than patients who underwent CTA/CTP alone (26 patients, 5%; and 2 patients, 5%, respectively; P = .7). CONCLUSIONS: Administration of a contrast-enhanced CT protocol involving CTA/CTP and conventional angiography in selected patients does not appear to increase the incidence of CIN.

Severity of leukoaraiosis correlates with clinical outcome after ischemic stroke.

BACKGROUND: Leukoaraiosis (LA) is closely associated with aging, a major determinant of clinical outcome after ischemic stroke. In this study we sought to identify whether LA, independent of advancing age, affects outcome after acute ischemic stroke. METHODS: LA volume was quantified in 240 patients with ischemic stroke and MRI within 24 hours of symptom onset. We explored the relationship between LA volume at admission and clinical outcome at 6 months, as assessed by the modified Rankin Scale (mRS). An ordinal logistic regression model was developed to analyze the independent effect of LA volume on clinical outcome. RESULTS: Bivariate analyses showed a significant correlation between LA volume and mRS at 6 months (r = 0.19, p = 0.003). Mean mRS was 1.7 +/- 1.8 among those in the lowest (< or =1.2 mL) and 2.5 +/- 1.9 in the highest (>9.9 mL) quartiles of LA volume (p = 0.01). The unfavorable prognostic effect of LA volume on clinical outcome was retained in the multivariable model (p = 0.002), which included age, gender, stroke risk factors (hypertension, diabetes mellitus, atrial fibrillation), previous history of brain infarction, admission plasma glucose level, admission NIH Stroke Scale score, IV rtPA treatment, and acute infarct volume on MRI as covariates. CONCLUSIONS: The volume of leukoaraiosis is a predictor of clinical outcome after ischemic stroke and this relationship persists after adjustment for important prognostic factors including age, initial stroke severity, and infarct volume.

Tumour-like mass lesion: an under-recognised presentation of primary angiitis of the central nervous system.

OBJECTIVE: To describe the occurrence of mass lesions (ML) in primary angiitis of the central nervous system (PACNS) and assess the utility of diagnostic testing and treatment. METHODS: We examined the case records of the Cleveland Clinic (CC), Massachusetts General Hospital (MGH), and the English language medical literature, for biopsy-proven PACNS cases presenting as a solitary ML. Relevant clinical variables were extracted and analysed with JMP software. RESULTS: We identified a total of 38 ML: eight of 202 (4.0%) patients with CC/MGH and 30 of 535 (5.6%) patients with PACNS identified from the medical literature. A higher percentage (13 of 45; 29%) was seen in the amyloid-related angiitis subset. Poorer outcomes were reported in the amyloid group, with five deaths. Of the non-amyloid group, better outcomes were seen in the group treated with corticosteroids and cyclophosphamide as compared with the group treated with corticosteroids alone. CONCLUSIONS: Although rare, PACNS should be considered in the differential diagnosis of ML; greater awareness of this manifestation may facilitate more prompt diagnosis and treatment. Biopsy evidence of angiitis is required for diagnosis; specimens should routinely be stained for amyloid. While excision of the lesion may be curative, aggressive immunosuppressive therapy is associated with favourable outcomes and may obviate the need for surgery.

CT angiography and diffusion-perfusion MR imaging in a patient with ipsilateral reversible cerebral vasoconstriction after carotid endarterectomy.

Reversible cerebral vasoconstriction syndromes (RCVS) typically affect the bilateral medium-sized intracerebral arteries and their branches. We describe a woman with RCVS restricted to the ipsilateral hemisphere after carotid endarterectomy. Serial CT angiography proved useful in documenting vasoconstriction. Perfusion MR imaging showed hypoperfusion in the deep watershed regions of the ipsilateral cerebral arteries but relatively normal perfusion in superficial cortical regions. Diffusion MR imaging showed progressive borderzone infarcts. These novel imaging findings provide insights into the pathophysiology of stroke in RCVS.

SSRI and statin use increases the risk for vasospasm after subarachnoid hemorrhage.

BACKGROUND: Use of medications with vasoconstrictive or vasodilatory effects can potentially affect the risk for vasospasm after aneurysmal subarachnoid hemorrhage (SAH). METHODS: Using International Classification of Diseases-9 diagnostic codes followed by medical record review, the authors identified 514 patients with SAH admitted between 1995 and 2003 who were evaluated for vasospasm between days 4 and 14. The authors determined risks for vasospasm, symptomatic vasospasm, and poor clinical outcomes in patients with documented pre-hemorrhagic use of calcium channel blockers, beta-receptor blockers, ACE inhibitors, aspirin, selective serotonin reuptake inhibitors (SSRIs), non-SSRI vasoactive antidepressants, or statins. RESULTS: Vasospasm developed in 62%, and symptomatic vasospasm in 29% of the cohort. On univariate analysis, the risk for all vasospasm tended to increase in patients taking SSRIs (p = 0.09) and statins (p = 0.05); SSRI use increased the risk for symptomatic vasospasm (p = 0.028). The Cochran-Armitage trend test showed that the proportion of patients taking SSRIs and statins increased significantly across three worsening categories (none, asymptomatic, symptomatic) of vasospasm. Logistic regression analysis showed that SSRI use tended to predict all vasospasm (O.R. 2.01 [0.91 to 4.45]), and predicted symptomatic vasospasm (O.R. 1.42 [1.06 to 4.33]). Statin exposure increased the risk for vasospasm (O.R. 2.75 [1.16 to 6.50]), perhaps from abrupt statin withdrawal (O.R. 2.54 [0.78 to 8.28]). Age < 50 years, Hunt-Hess grade 4 or 5, and Fisher Group 3 independently predicted all vasospasm, symptomatic vasospasm, poor discharge clinical status, and death. CONCLUSION: Selective serotonin reuptake inhibitor and statin users have a higher risk for subarachnoid hemorrhage-related vasospasm. Whether the underlying disease indication, direct actions, or rebound effects from abrupt drug withdrawal account for the associated risk warrants further investigation.

Cerebral vasoconstriction and stroke after use of serotonergic drugs.

Serotonin (5-hydroxytryptamine) is a potent vasoconstrictor amine. The authors report three patients who developed thunderclap headache, reversible cerebral arterial vasoconstriction, and ischemic strokes (i.e., the Call-Fleming syndrome). The only cause for vasoconstriction was recent exposure to serotonergic drugs in all patients, and to pseudoephedrine in one patient. These cases, and the literature, suggest that the use of serotonin-enhancing drugs can precipitate a cerebrovascular syndrome due to reversible, multifocal arterial narrowing.

Drug delivery to damaged brain.

Drug delivery to the brain poses unique challenges. Specialized anatomic and physiological features of the cerebrovasculature and cerebral tissue fluids result in barriers which significantly restrict delivery of a wide range of possible therapeutic agents. In addition to these normal restrictions to brain drug delivery, pathophysiological features and sequelae of acute brain injury will also impact upon the efficiency of drug delivery. This review is focused on acutely damaged brain that occurs after stroke and trauma. Pathophysiological events that may influence drug delivery include blood-brain barrier disruptions, blood flow alterations, edema and increased intracranial pressure, metabolic perturbations, and altered profiles of gene expression and protein synthesis. Careful consideration of these obstacles will provide a framework for further research into the optimization of drug delivery strategies into damaged brain. Without a rigorous assessment of these issues, it may not be possible to translate our mechanistic understanding of acute brain injury into successful clinical therapies.

Protective effects of pamiteplase, a modified t-PA, in a rat model of embolic stroke.

The effects of alteplase (tissue plasminogen activator, t-PA) and pamiteplase (a modified t-PA with longer half-life and increased potency) were compared in a clinically relevant model of embolic stroke. Rats were treated with pamiteplase (0.5 mg/kg or 1 mg/kg bolus), alteplase (10 mg/kg infusion) or normal saline. Pamiteplase (1 mg/kg) was as effective as alteplase in reducing 24 h brain infarct volumes, neurological deficit scores and residual clot grades. Cerebral blood flow recovery at 30 min after thrombolytic treatment was partial and did not correlate with 24 h infarct volumes or neurological deficits. However, there was good correlation between 24 h residual clot grades and infarct volumes, suggesting a delayed timeframe for pamiteplase- and alteplase-induced reperfusion.

Cerebral venous sinus thrombosis associated with hepatic cirrhosis.

Cerebral venous sinus thrombosis is not a recognized complication of end-stage liver disease. A case of sagittal sinus thrombosis in a 44-year-old male with end-stage hepatic cirrhosis is described. Recurrent seizures were the only manifestation. Work-up revealed severe deficiency of protein C, protein S, and antithrombin III. He was treated with low molecular weight heparin and underwent an orthotopic liver transplant after three months. Follow-up helical CT venogram showed resolution of the sinus thrombosis.

High level functional engraftment of severe combined immunodeficient mice with human peripheral blood lymphocytes following pretreatment with radiation and anti-asialo GM1.

The severe combined immunodeficient (SCID) mouse engrafted with human peripheral blood lymphocytes (PBLs) is a potentially useful model for the study of cancer immunotherapy. For this application, rapid, consistent, and high level engraftment of SCID mice with functional human cytotoxic effector cells is necessary. To date, short term human lymphoid cell engraftment in SCID mice has generally been low and variable. Further, most of the human cells detected within the first 30 days are found in the peritoneal cavity. The purpose of the present study was to improve short term reconstitution of human PBLs in the SCID mouse. When untreated SCID mice were injected with human PBLs, the mean level of CD3+ cells in the spleens was < 5% on days 6-32 after injection, as determined by flow cytometry (FCM). Depletion of SCID mouse natural killer (NK) cells with anti-asialo GM1 only marginally improved short term reconstitution with human CD3+ cells. Preirradiation of SCID mice with 3 Gy improved reconstitution to over 16% CD3+ cells on days 12-14 following engraftment. However, the combination of pretreatment with anti-asialo GM1 plus radiation, significantly increased the mean percentage of human CD3+ cells in the spleen to 40% within 2 weeks following injection of PBLs. Human T cells positive for CD4, CD8, TcR alpha beta, and TcR gamma delta, and human NK and B cells were detected in the spleens of irradiated plus anti-asialo GM1 pretreated SCID mice. The presence of human lymphoid cells was confirmed by immunohistologic staining. The human immune cells in these mice were shown to be functional by the in vivo demonstration of an appropriate secondary immune response to the injection of tetanus toxoid and by an in vivo proliferative response to phytohemagglutinin. Human NK cells could be found in the spleens and peripheral blood of irradiated plus anti-asialo GM1 pretreated mice. These cells were also shown to be competent by their ability to lyse the human NK sensitive tumor targets K562 and MOLT-4 in 51Cr release assays. Thus, pretreatment of SCID mice with radiation plus anti-asialo GM1 significantly improves short term human PBL engraftment and provides a potentially useful model for the study of cancer immunotherapy.