Radiomic-based nonlinear supervised learning classifiers on non-contrast CT to predict functional prognosis in patients with spontaneous intracerebral hematoma.

PURPOSE: To evaluate if nonlinear supervised learning classifiers based on non-contrast CT can predict functional prognosis at discharge in patients with spontaneous intracerebral hematoma. METHODS: Retrospective, single-center, observational analysis of patients with a diagnosis of spontaneous intracerebral hematoma confirmed by non-contrast CT between January 2016 and April 2018. Patients with HIE > 18 years and with TCCSC performed within the first 24 h of symptom onset were included. Patients with secondary spontaneous intracerebral hematoma and in whom radiomic variables were not available were excluded. Clinical, demographic and admission variables were collected. Patients were classified according to the Modified Rankin Scale (mRS) at discharge into good (mRS 0-2) and poor prognosis (mRS 3-6). After manual segmentation of each spontaneous intracerebral hematoma, the radiomics variables were obtained. The sample was divided into a training and testing cohort and a validation cohort (70-30% respectively). Different methods of variable selection and dimensionality reduction were used, and different algorithms were used for model construction. Stratified 10-fold cross-validation were performed on the training and testing cohort and the mean area under the curve (AUC) were calculated. Once the models were trained, the sensitivity of each was calculated to predict functional prognosis at discharge in the validation cohort. RESULTS: 105 patients with spontaneous intracerebral hematoma were analyzed. 105 radiomic variables were evaluated for each patient. P-SVM, KNN-E and RF-10 algorithms, in combination with the ANOVA variable selection method, were the best performing classifiers in the training and testing cohort (AUC 0.798, 0.752 and 0.742 respectively). The predictions of these models, in the validation cohort, had a sensitivity of 0.897 (0.778-1;95%CI), with a false-negative rate of 0% for predicting poor functional prognosis at discharge. CONCLUSION: The use of radiomics-based nonlinear supervised learning classifiers are a promising diagnostic tool for predicting functional outcome at discharge in HIE patients, with a low false negative rate, although larger and balanced samples are still needed to develop and improve their performance.

Optimizing the Definition of Ischemic Core in CT Perfusion: Influence of Infarct Growth and Tissue-Specific Thresholds.

BACKGROUND AND PURPOSE: CTP allows estimating ischemic core in patients with acute stroke. However, these estimations have limited accuracy compared with MR imaging. We studied the effect of applying WM- and GM-specific thresholds and analyzed the infarct growth from baseline imaging to reperfusion. MATERIALS AND METHODS: This was a single-center cohort of consecutive patients (n = 113) with witnessed strokes due to proximal carotid territory occlusions with baseline CT perfusion, complete reperfusion, and follow-up DWI. We segmented GM and WM, coregistered CTP with DWI, and compared the accuracy of the different predictions for each voxel on DWI through receiver operating characteristic analysis. We assessed the yield of different relative CBF thresholds to predict the final infarct volume and an estimated infarct growth-corrected volume (subtracting the infarct growth from baseline imaging to complete reperfusion) for a single relative CBF threshold and GM- and WM-specific thresholds. RESULTS: The fixed threshold underestimated lesions in GM and overestimated them in WM. Double GM- and WM-specific thresholds of relative CBF were superior to fixed thresholds in predicting infarcted voxels. The closest estimations of the infarct on DWI were based on a relative CBF of 25% for a single threshold, 35% for GM, and 20% for WM, and they decreased when correcting for infarct growth: 20% for a single threshold, 25% for GM, and 15% for WM. The combination of 25% for GM and 15% for WM yielded the best prediction. CONCLUSIONS: GM- and WM-specific thresholds result in different estimations of ischemic core in CTP and increase the global accuracy. More restrictive thresholds better estimate the actual extent of the infarcted tissue.

Synthetic MRI in subarachnoid haemorrhage.

AIM: To evaluate the reliability of synthetic magnetic resonance imaging (SyMRI) for detecting complications associated with subarachnoid haemorrhage (SAH), such as ischaemic lesions, hydrocephalus, or bleeding complications. MATERIALS AND METHODS: Twenty patients with SAH, who underwent a conventional brain MRI and a SyMRI on a 3 T MRI machine. Comparable conventional and synthetic T2-weighted fluid attenuated inversion recovery (FLAIR) images were acquired. The presence of ischaemic lesions, hydrocephalus, extra-axial blood collections as well as the volumes of grey matter (GMv), white matter (WMv), and cerebrospinal (CSFv) were compared. The acquisition times of both sequences was also analysed. RESULTS: The concordance between the two techniques was excellent for the detection of ischaemic lesions and extra-axial collections (kappa = 0.80 and 0.88 respectively) and good for the detection of hydrocephalus (kappa = 0.69). No significant differences were detected in the number of ischaemic lesions (p=0.31) or in the Evans index (p=0.11). The WMv and CSFv measures were also similar (p=0.18 and p=0.94, respectively), as well as the volume of ischaemic lesions (p=0.79). Compared to conventional MRI, the SyMRI acquisition time was shorter regardless of the number of sections (32% and 6% time reduction for 4 or 3 mm section thickness, respectively). CONCLUSIONS: SyMRI allows the detection of potential complications of SAH in a similar way to conventional MRI with a shorter acquisition time.

Stroke units could be a valid alternative to intensive care units for patients with low-grade aneurysmal subarachnoid haemorrhage.

BACKGROUND AND PURPOSE: According to current guidelines, patients with aneurysmal subarachnoid haemorrhage (aSAH) are mostly managed in intensive care units (ICUs) regardless of baseline severity. We aimed to assess the prognostic and economic implications of initial admission of patients with low-grade aSAH into a stroke unit (SU) compared to initial ICU admission. METHODS: We reviewed prospectively registered data from consecutive aSAH patients with a World Federation of Neurosurgery Societies grade <3, admitted to our Comprehensive Stroke Centre between April 2013 and September 2018. Clinical and radiological baseline traits, in-hospital complications, length of stay (LOS) and poor outcome at 90 days (modified Rankin Scale score > 2) were compared between the ICU and SU groups in the whole population and in a propensity-score-matched cohort. RESULTS: Of 131 patients, 74 (56%) were initially admitted to the ICU and 57 (44%) to the SU. In-hospital complication rates were similar in the ICU and SU groups and included rebleeding (10% vs. 7%; P = 0.757), angiographic vasospasm (61% vs. 60%; P = 0.893), delayed cerebral ischaemia (12% vs. 12%; P = 0.984), pneumonia (6% vs. 4%; P = 0.697) and death (10% vs. 5%; P = 0.512). LOS did not differ between groups (median [interquartile range] 22 [16-30] vs. 19 [14-26] days; P = 0.160). In adjusted multivariate models, the location of initial admission was not associated with long-term poor outcome either in the whole population (odds ratio [OR] 1.16, 95% confidence interval [CI] 0.32-4.19; P = 0.825) or in the matched cohort (OR 0.98, 95% CI 0.24-4.06; P = 0.974). CONCLUSIONS: A dedicated SU, with care from a multidisciplinary team, might be an optimal alternative to ICU for initial admission of patients with low-risk aSAH.

Leukoaraiosis May Confound the Interpretation of CT Perfusion in Patients Treated with Mechanical Thrombectomy for Acute Ischemic Stroke.

BACKGROUND AND PURPOSE: Leukoaraiosis frequently coexists in patients with acute stroke. We studied whether leukoaraiosis could confound the interpretation of CTP findings in patients treated with mechanical thrombectomy. MATERIALS AND METHODS: We analyzed 236 patients with stroke treated with mechanical thrombectomy and studied with CTP, of whom 127 (53.8%) achieved complete reperfusion. Periventricular white matter hyperintensities on MR imaging and hypodensities on NCCT were assessed through the Fazekas score. CTP-predicted nonviable tissue was defined as relative CBF <30%, and final infarct volume was quantified in DWI. We estimated mean MTT, CBV, and CBF in the asymptomatic hemisphere. In patients achieving complete reperfusion, we assessed the accuracy of nonviable tissue to predict final infarct volume using the intraclass correlation coefficient across periventricular hyperintensity/hypodensity Fazekas scores and variable relative CBF cutoffs. RESULTS: MTT was longer (Spearman ρ = 0.279, P < .001) and CBF was lower (ρ = -0.263, P < .001) as the periventricular hyperintensity Fazekas score increased, while CBV was similar across groups (ρ = -0.043, P = .513). In the subgroup of patients achieving complete reperfusion, nonviable tissue-final infarct volume reliability was excellent in patients with periventricular hyperintensity Fazekas score grade 0 (intraclass correlation coefficient, 0.900; 95% CI, 0.805-0.950), fair in patients with periventricular hyperintensity Fazekas scores 1 (intraclass correlation coefficient, 0.569; 95% CI, 0.327-0.741) and 2 (intraclass correlation coefficient, 0.444; 95% CI, 0.165-0.657), and poor in patients with periventricular hyperintensity Fazekas score 3 (intraclass correlation coefficient, 0.310; 95% CI, -0.359-0.769). The most accurate cutoffs were relative CBF <30% for periventricular hyperintensity Fazekas score grades 0 and 1, relative CBF <25% for periventricular hyperintensity Fazekas score 2, and relative CBF <20% for periventricular hyperintensity Fazekas score 3. The reliability analysis according to periventricular hypodensity Fazekas score grades on NCCT was similar to that in follow-up MR imaging. CONCLUSIONS: In patients with stroke, the presence of leukoaraiosis confounds the interpretation of CTP despite proper adjustment of CBF thresholds.

The neurologist's approach to cerebral infarct and transient ischaemic attack in patients receiving anticoagulant treatment for non-valvular atrial fibrillation: ANITA-FA study.

BACKGROUND AND PURPOSE: The aim of this study was to describe the clinical and epidemiological characteristics of acute ischaemic stroke (AIS) in patients with atrial fibrillation (AF) previously treated with oral anticoagulants (OACs) according to the type of OAC prescribed. Also, to analyze the outcomes of the patients and the therapeutic approach adopted by the neurologist in the acute phase and for secondary prevention. METHODS: We performed a multicenter, observational study based on prospective registries. We included patients with AF treated with OACs admitted for AIS over a 1-year period. Detailed clinical data and functional outcome at 3 months (modified Rankin Scale score) were collected. Patients were divided into two groups according to their pre-AIS anticoagulant therapy: vitamin K antagonists (AIS-VKA) and direct-acting OACs (AIS-DOAC). RESULTS: We recruited 1240 patients (80.4% AIS-VKA and 19.6% AIS-DOAC). In the AIS-DOAC group, transient ischaemic attack was more frequent (18.1% vs. 10.8%; P = 0.001), symptomatic hemorrhagic transformation was less frequent (1.6% vs. 4.6%; P = 0.035) and hospital stay was shorter (median 6 vs. 7 days; P = 0.03). Intravenous thrombolysis was more commonly used in AIS-VKA (9.2% vs. 1.6%; P < 0.001). There were no differences between the groups with respect to mechanical thrombectomy, mortality and modified Rankin Scale score at 3 months. At 3 months, 54% of patients required a DOAC as antithrombotic treatment for secondary prevention. CONCLUSIONS: Patients with AF treated with DOACs who experienced AIS more frequently had transient symptoms (transient ischaemic attack), less symptomatic hemorrhagic transformation and a shorter mean stay than those treated with VKAs. Most patients who had been previously anticoagulated with AIS received long-term treatment with DOACs.