Sepsis and delayed cerebral ischemia are associated and have a cumulative effect on poor functional outcome in aneurysmal subarachnoid hemorrhage.

Objective: Although sepsis and delayed cerebral ischemia (DCI) are severe complications in patients with aneurysmal subarachnoid hemorrhage (aSAH) and share pathophysiological features, their interrelation and additive effect on functional outcome is uncertain. We investigated the association between sepsis and DCI and their cumulative effect on functional outcome in patients with aSAH using current sepsis-3 definition. Methods: Patients admitted to our hospital between 11/2014 and 11/2018 for aSAH were retrospectively analyzed. The main explanatory variable was sepsis, diagnosed using sepsis-3 criteria. Endpoints were DCI and functional outcome at hospital discharge (modified Rankin Scale (mRS) 0-3 vs. 4-6). Propensity score matching (PSM) and multivariable logistic regressions were performed. Results: Of 238 patients with aSAH, 55 (23.1%) developed sepsis and 74 (31.1%) DCI. After PSM, aSAH patients with sepsis displayed significantly worse functional outcome (p < 0.01) and longer ICU stay (p = 0.046). Sepsis was independently associated with DCI (OR = 2.46, 95%CI: 1.28-4.72, p < 0.01). However, after exclusion of patients who developed sepsis before (OR = 1.59, 95%CI: 0.78-3.24, p = 0.21) or after DCI (OR = 0.85, 95%CI: 0.37-1.95, p = 0.70) this statistical association did not remain. Good functional outcome gradually decreased from 56.3% (76/135) in patients with neither sepsis nor DCI, to 43.8% (21/48) in those with no sepsis but DCI, to 34.5% (10/29) with sepsis but no DCI and to 7.7% (2/26) in patients with both sepsis and DCI. Conclusion: Our study demonstrates a strong association between sepsis, DCI and functional outcome in patients with aSAH and suggests a complex interplay resulting in a cumulative effect towards poor functional outcome, which warrants further studies.

Prediction of tissue outcome in acute ischemic stroke based on single-phase CT angiography at admission.

Introduction: In acute ischemic stroke, prediction of the tissue outcome after reperfusion can be used to identify patients that might benefit from mechanical thrombectomy (MT). The aim of this work was to develop a deep learning model that can predict the follow-up infarct location and extent exclusively based on acute single-phase computed tomography angiography (CTA) datasets. In comparison to CT perfusion (CTP), CTA imaging is more widely available, less prone to artifacts, and the established standard of care in acute stroke imaging protocols. Furthermore, recent RCTs have shown that also patients with large established infarctions benefit from MT, which might not have been selected for MT based on CTP core/penumbra mismatch analysis. Methods: All patients with acute large vessel occlusion of the anterior circulation treated at our institution between 12/2015 and 12/2020 were screened (N = 404) and 238 patients undergoing MT with successful reperfusion were included for final analysis. Ground truth infarct lesions were segmented on 24 h follow-up CT scans. Pre-processed CTA images were used as input for a U-Net-based convolutional neural network trained for lesion prediction, enhanced with a spatial and channel-wise squeeze-and-excitation block. Post-processing was applied to remove small predicted lesion components. The model was evaluated using a 5-fold cross-validation and a separate test set with Dice similarity coefficient (DSC) as the primary metric and average volume error as the secondary metric. Results: The mean ± standard deviation test set DSC over all folds after post-processing was 0.35 ± 0.2 and the mean test set average volume error was 11.5 mL. The performance was relatively uniform across models with the best model according to the DSC achieved a score of 0.37 ± 0.2 after post-processing and the best model in terms of average volume error yielded 3.9 mL. Conclusion: 24 h follow-up infarct prediction using acute CTA imaging exclusively is feasible with DSC measures comparable to results of CTP-based algorithms reported in other studies. The proposed method might pave the way to a wider acceptance, feasibility, and applicability of follow-up infarct prediction based on artificial intelligence.

Endovascular Thrombectomy for Large Ischemic Strokes with ASPECTS 0-2: a Meta-analysis of Randomized Controlled Trials.

PURPOSE: Randomized controlled trials (RCTs) demonstrated a treatment effect of endovascular thrombectomy in acute ischemic stroke with large infarct, commonly defined as an Alberta Stroke Program Early CT Score (ASPECTS) of 3-5. However, data on endovascular thrombectomy in patients with very low ASPECTS of 0-2 remain scarce. METHODS: We conducted a systematic review and meta-analysis of RCTs comparing endovascular thrombectomy versus medical treatment alone in acute ischemic anterior circulation stroke with very large infarct, defined as ASPECTS of 0-2. The primary outcome was the shift toward better functional outcomes on the 90-day modified Rankin Scale (mRS). Random effects meta-analysis was performed using the generic inverse variance method. RESULTS: Literature research identified four RCTs which evaluated the treatment effect of endovascular thrombectomy for large infarcts and provided a subgroup analysis of the mRS shift in patients with ASPECTS of 0-2. The pooled analysis showed a significant shift toward better 90-day mRS scores in favor of endovascular thrombectomy (pooled odds ratio, 1.62, 95% confidence interval, 1.29-2.04, P < 0.001). CONCLUSION: This meta-analysis suggests a treatment effect of endovascular thrombectomy in specific patients with very low ASPECTS of 0-2, challenging the use of ASPECTS for treatment selection in acute ischemic stroke due to large vessel occlusion. An individual patient meta-analysis of RCTs would strengthen evidence in the treatment of patients with ASPECTS of 0-2.

Association Between Recanalization Attempts and Functional Outcome After Thrombectomy for Large Ischemic Stroke.

BACKGROUND: Recently, 3 randomized controlled trials provided high-level evidence that patients with large ischemic stroke achieved better functional outcomes after endovascular therapy than with medical care alone. We aimed to investigate whether the clinical benefit of endovascular therapy is associated with the number of recanalization attempts in extensive baseline infarction. METHODS: This retrospective multicenter study enrolled patients from the German Stroke Registry who underwent endovascular therapy for anterior circulation large vessel occlusion between 2015 and 2021. Large ischemic stroke was defined as an Alberta Stroke Program Early Computed Tomography Score of 3 to 5. The study cohort was divided into patients with unsuccessful reperfusion (Thrombolysis in Cerebral Infarction score, 0-2a) and successful reperfusion (Thrombolysis in Cerebral Infarction score, 2b/3) at attempts 1, 2, 3, or ≥4. The primary outcome was favorable functional outcome defined as modified Rankin Scale score of 0 to 3 at 90 days. Safety outcomes were symptomatic intracranial hemorrhage after 24 hours and death within 90 days. Multivariable logistic regression was used to identify independent determinants of primary and secondary outcomes. RESULTS: A total of 348 patients met the inclusion criteria. Successful reperfusion was observed in 83.3% and favorable functional outcomes in 36.2%. Successful reperfusion at attempts 1 (adjusted odds ratio, 5.97 [95% CI, 1.71-24.43]; P=0.008) and 2 (adjusted odds ratio, 6.32 [95% CI, 1.73-26.92]; P=0.008) increased the odds of favorable functional outcome, whereas success at attempts 3 or ≥4 did not. Patients with >2 attempts showed higher rates of symptomatic intracranial hemorrhage (12.8% versus 6.5%; P=0.046). Successful reperfusion at any attempt lowered the odds of death compared with unsuccessful reperfusion. CONCLUSIONS: In patients with large vessel occlusion and Alberta Stroke Program Early Computed Tomography Score of 3 to 5, the clinical benefit of endovascular therapy was linked to the number of recanalization attempts required for successful reperfusion. Our findings encourage to perform at least 2 recanalization attempts to seek for successful reperfusion in large ischemic strokes, while >2 attempts should follow a careful risk-benefit assessment in these highly affected patients. REGISTRATION: URL: https://www. CLINICALTRIALS: gov; Unique identifier: NCT03356392.

Individual glioblastoma cells harbor both proliferative and invasive capabilities during tumor progression.

BACKGROUND: Glioblastomas are characterized by aggressive and infiltrative growth, and by striking heterogeneity. The aim of this study was to investigate whether tumor cell proliferation and invasion are interrelated, or rather distinct features of different cell populations. METHODS: Tumor cell invasion and proliferation were longitudinally determined in real-time using 3D in vivo 2-photon laser scanning microscopy over weeks. Glioblastoma cells expressed fluorescent markers that permitted the identification of their mitotic history or their cycling versus non-cycling cell state. RESULTS: Live reporter systems were established that allowed us to dynamically determine the invasive behavior, and previous or actual proliferation of distinct glioblastoma cells, in different tumor regions and disease stages over time. Particularly invasive tumor cells that migrated far away from the main tumor mass, when followed over weeks, had a history of marked proliferation and maintained their proliferative capacity during brain colonization. Infiltrating cells showed fewer connections to the multicellular tumor cell network, a typical feature of gliomas. Once tumor cells colonized a new brain region, their phenotype progressively transitioned into tumor microtube-rich, interconnected, slower-cycling glioblastoma cells. Analysis of resected human glioblastomas confirmed a higher proliferative potential of tumor cells from the invasion zone. CONCLUSIONS: The detection of glioblastoma cells that harbor both particularly high proliferative and invasive capabilities during brain tumor progression provides valuable insights into the interrelatedness of proliferation and migration-2 central traits of malignancy in glioma. This contributes to our understanding of how the brain is efficiently colonized in this disease.

PSEN1/SLC20A2 double mutation causes early-onset Alzheimer's disease and primary familial brain calcification co-morbidity.

Primary familial brain calcification (PFBC; formerly Fahr's disease) and early-onset Alzheimer's disease (EOAD) may share partially overlapping pathogenic principles. Although the heterozygous loss-of-function mutation c.1523 + 1G > T in the PFBC-linked gene SLC20A2 was detected in a patient with asymmetric tremor, early-onset dementia, and brain calcifications, CSF ß-amyloid parameters and FBB-PET suggested cortical ß-amyloid pathology. Genetic re-analysis of exome sequences revealed the probably pathogenic missense mutation c.235G > A/p.A79T in PSEN1. The SLC20A2 mutation segregated with mild calcifications in two children younger than 30 years. We thus describe the stochastically extremely unlikely co-morbidity of genetic PFBC and genetic EOAD. The clinical syndromes pointed to additive rather than synergistic effects of the two mutations. MRI data revealed the formation of PFBC calcifications decades before the probable onset of the disease. Our report furthermore exemplifies the value of neuropsychology and amyloid PET for differential diagnosis.

Lightweight Visual Transformers Outperform Convolutional Neural Networks for Gram-Stained Image Classification: An Empirical Study.

We aimed to automate Gram-stain analysis to speed up the detection of bacterial strains in patients suffering from infections. We performed comparative analyses of visual transformers (VT) using various configurations including model size (small vs. large), training epochs (1 vs. 100), and quantization schemes (tensor- or channel-wise) using float32 or int8 on publicly available (DIBaS, n = 660) and locally compiled (n = 8500) datasets. Six VT models (BEiT, DeiT, MobileViT, PoolFormer, Swin and ViT) were evaluated and compared to two convolutional neural networks (CNN), ResNet and ConvNeXT. The overall overview of performances including accuracy, inference time and model size was also visualized. Frames per second (FPS) of small models consistently surpassed their large counterparts by a factor of 1-2×. DeiT small was the fastest VT in int8 configuration (6.0 FPS). In conclusion, VTs consistently outperformed CNNs for Gram-stain classification in most settings even on smaller datasets.

Myelofibrosis progression grading based on type I and type III collagen and fibrillin 1 expression boosted by whole slide image analysis.

AIMS: The progression of primary myelofibrosis is characterised by ongoing extracellular matrix deposition graded based on 'reticulin' and 'collagen' fibrosis, as revealed by Gomori's silver impregnation. Here we studied the expression of the major extracellular matrix proteins of fibrosis in relation to diagnostic silver grading supported by image analysis. METHODS AND RESULTS: By using automated immunohistochemistry, in this study we demonstrate that the expression of both types I and III collagens and fibrillin 1 by bone marrow stromal cells can reveal the extracellular matrix scaffolding in line with myelofibrosis progression as classified by silver grading. 'Reticulin' fibrosis indicated by type III collagen expression and 'collagen' fibrosis featured by type I collagen expression were parallel, rather than sequential, events. This is line with the proposed role of type III collagen in regulating type I collagen fibrillogenesis. The uniformly strong fibrillin 1 immune signals offered the best inter-rater agreements and the highest statistical correlations with silver grading of the three markers, which was robustly confirmed by automated whole slide digital image analysis using a machine learning-based algorithm. The progressive up-regulation of fibrillin 1 during myelofibrosis may result from a negative feedback loop as fibrillin microfibrils sequester TGF-ß, the major promoter of fibrosis. This can also reduce TGF-ß-induced RANKL levels, which would stimulate osteoclastogenesis and thus can support osteosclerosis in advanced myelofibrosis. CONCLUSIONS: Through the in-situ detection of these extracellular matrix proteins, our results verify the molecular pathobiology of fibrosis during myelofibrosis progression. In particular, fibrillin 1 immunohistochemistry, with or without image analysis, can complement diagnostic silver grading at decent cell morphology.

Rapid Convolutional Neural Networks for Gram-Stained Image Classification at Inference Time on Mobile Devices: Empirical Study from Transfer Learning to Optimization.

Despite the emergence of mobile health and the success of deep learning (DL), deploying production-ready DL models to resource-limited devices remains challenging. Especially, during inference time, the speed of DL models becomes relevant. We aimed to accelerate inference time for Gram-stained analysis, which is a tedious and manual task involving microorganism detection on whole slide images. Three DL models were optimized in three steps: transfer learning, pruning and quantization and then evaluated on two Android smartphones. Most convolutional layers (≥80%) had to be retrained for adaptation to the Gram-stained classification task. The combination of pruning and quantization demonstrated its utility to reduce the model size and inference time without compromising model quality. Pruning mainly contributed to model size reduction by 15×, while quantization reduced inference time by 3× and decreased model size by 4×. The combination of two reduced the baseline model by an overall factor of 46×. Optimized models were smaller than 6 MB and were able to process one image in <0.6 s on a Galaxy S10. Our findings demonstrate that methods for model compression are highly relevant for the successful deployment of DL solutions to resource-limited devices.

Influence of neurovascular anatomy on perforation site in different mouse strains using the filament perforation model for induction of subarachnoid hemorrhage.

BACKGROUND: Filament perforation is a widely-used method to induce subarachnoid hemorrhage (SAH) in mice. Whereas the perforation site has been assumed to be in the branching of middle cerebral artery (MCA) and anterior cerebral artery (ACA), we recently observed more proximal perforations. METHODS: Filament perforation was performed in CD1- (n = 10) and C57Bl/6N-mice (n = 9) ex vivo. The filament was left in place and the perforation site was microscopically assessed. Digital subtraction angiography (DSA) was performed in CD1- (n = 9) and C57Bl/6J-mice (n = 29) and anatomical differences of the internal carotid artery (ICA) were determined. RESULTS: Whereas in C57Bl/6N-mice perforation occurred in the proximal intracranial ICA in 89% (n = 8), in CD1-mice the perforation site was in the proximal ICA in 50% (n = 5), in the branching between MCA and ACA in 40% (n = 4), and in the proximal ACA in 10% (n = 1). DSA revealed a stronger angulation (p<0.001) of the ICA in CD1-mice (163.5±2.81°) compared to C57Bl/6J-mice (124.5±5.49°). Body weight and ICA-angle showed no significant correlation in C57Bl/6J- (r = -0.06, pweight/angle = 0.757) and CD1-mice (r = -0.468, pweight/angle = 0.242). CONCLUSION: Filament perforation in mice occurs not only at the hitherto presumed branching between MCA and ACA, but seems to depend on mouse strain and anatomy as the proximal intracranial ICA may also be perforated frequently.

Comparative Analyses of the Impact of Different Criteria for Sepsis Diagnosis on Outcome in Patients with Spontaneous Subarachnoid Hemorrhage.

Data on sepsis in patients with a subarachnoid hemorrhage (SAH) are scarce. We assessed the impact of different sepsis criteria on the outcome in an SAH cohort. Adult patients admitted to our ICU with a spontaneous SAH between 11/2014 and 11/2018 were retrospectively included. In patients developing an infection, different criteria for sepsis diagnosis (Sepsis-1, Sepsis-3_original, Sepsis-3_modified accounting for SAH-specific therapy, alternative sepsis criteria compiled of consensus conferences) were applied and their impact on functional outcome using the modified Rankin Scale (mRS) on hospital discharge and in-hospital mortality was evaluated. Of 270 SAH patients, 129 (48%) developed an infection. Depending on the underlying criteria, the incidence of sepsis and septic shock ranged between 21-46% and 9-39%. In multivariate logistic regression, the Sepsis-1 criteria were not associated with the outcome. The Sepsis-3 criteria were not associated with the functional outcome, but in shock with mortality. Alternative sepsis criteria were associated with mortality for sepsis and in shock with mortality and the functional outcome. While Sepsis-1 criteria were irrelevant for the outcome in SAH patients, septic shock, according to the Sepsis-3 criteria, adversely impacted survival. This impact was higher for the modified Sepsis-3 criteria, accounting for SAH-specific treatment. Modified Sepsis-3 and alternative sepsis criteria diagnosed septic conditions of a higher relevance for outcomes in patients with an SAH.

Transfer learning for medical image classification: a literature review.

BACKGROUND: Transfer learning (TL) with convolutional neural networks aims to improve performances on a new task by leveraging the knowledge of similar tasks learned in advance. It has made a major contribution to medical image analysis as it overcomes the data scarcity problem as well as it saves time and hardware resources. However, transfer learning has been arbitrarily configured in the majority of studies. This review paper attempts to provide guidance for selecting a model and TL approaches for the medical image classification task. METHODS: 425 peer-reviewed articles were retrieved from two databases, PubMed and Web of Science, published in English, up until December 31, 2020. Articles were assessed by two independent reviewers, with the aid of a third reviewer in the case of discrepancies. We followed the PRISMA guidelines for the paper selection and 121 studies were regarded as eligible for the scope of this review. We investigated articles focused on selecting backbone models and TL approaches including feature extractor, feature extractor hybrid, fine-tuning and fine-tuning from scratch. RESULTS: The majority of studies (n = 57) empirically evaluated multiple models followed by deep models (n = 33) and shallow (n = 24) models. Inception, one of the deep models, was the most employed in literature (n = 26). With respect to the TL, the majority of studies (n = 46) empirically benchmarked multiple approaches to identify the optimal configuration. The rest of the studies applied only a single approach for which feature extractor (n = 38) and fine-tuning from scratch (n = 27) were the two most favored approaches. Only a few studies applied feature extractor hybrid (n = 7) and fine-tuning (n = 3) with pretrained models. CONCLUSION: The investigated studies demonstrated the efficacy of transfer learning despite the data scarcity. We encourage data scientists and practitioners to use deep models (e.g. ResNet or Inception) as feature extractors, which can save computational costs and time without degrading the predictive power.

Correlations Between the Expression of Stromal Cell Activation Related Biomarkers, L-NGFR, Phospho-ERK1-2 and CXCL12, and Primary Myelofibrosis Progression.

In myelofibrosis, pathologically enhanced extracellular matrix production due to aberrant cytokine signalling and clonal megakaryocyte functions result(s) in impaired hemopoiesis. Disease progression is still determined by detecting reticulin and collagen fibrosis with Gomori's silver impregnation. Here, we tested whether the expression growth related biomarkers L-NGFR/CD271, phospho-ERK1-2 and CXCL12 can be linked to the functional activation of bone marrow stromal cells during primary myelofibrosis progression. Immunoscores for all tested biomarkers showed varying strength of positive statistical correlation with the silver impregnation based myelofibrosis grades. The intimate relationship between spindle shaped stromal cells positive for all three markers and aberrant megakaryocytes was likely to reflect their functional cooperation. L-NGFR reaction was restricted to bone marrow stromal cells and revealed the whole length of their processes. Also, L-NGFR positive cells showed the most intersections, the best statistical correlations with myelofibrosis grades and the strongest interrater agreements. CXCL12 reaction highlighted stromal cell bodies and a weak extracellular staining in line with its constitutive release. Phospho-ERK1-2 reaction showed a similar pattern to CXCL12 in stromal cells with an additional nuclear staining in agreement with its role as a transcription factor. Both p-ERK1-2 and CXCL12 were also expressed at a moderate level in sinus endothelial cells. Connexin 43 gap junction communication channels, known to be required for CXCL12 release to maintain stem cell niche, were also expressed progressively in the myelofibrotic stromal network as a support of compartmental functions. Our results suggest that, diverse growth related pathways are activated in the functionally coupled bone marrow stromal cells during myelofibrosis progression. L-NGFR expression can be a useful biological marker of stromal cell activation which deserves diagnostic consideration for complementing Gomori's silver impregnation.

Cell Cycle Regulatory Protein Expression in Multinucleated Giant Cells of Giant Cell Tumor of Bone: do They Proliferate?

Cells of the monocyte macrophage lineage form multinucleated giant cells (GCs) by fusion, which may express some cell cycle markers. By using a comprehensive marker set, here we looked for potential replication activities in GCs, and investigated whether these have diagnostic or clinical relevance in giant cell tumor of bone (GCTB). GC rich regions of 10 primary and 10 first recurrence GCTB cases were tested using immunohistochemistry in tissue microarrays. The nuclear positivity rate of the general proliferation marker, replication licensing, G1/S-phase, S/G2/M-phase, mitosis promoter, and cyclin dependent kinase (CDK) inhibitor reactions was analyzed in GCs. Concerning Ki67, moderate SP6 reaction was seen in many GC nuclei, while B56 and Mib1 positivity was rare, but the latter could be linked to more aggressive (p = 0.012) phenotype. Regular MCM6 reaction, as opposed to uncommon MCM2, suggested an initial DNA unwinding. Early replication course in GCs was also supported by widely detecting CDK4 and cyclin E, for the first time, and confirming cyclin D1 upregulation. However, post-G1-phase markers CDK2, cyclin A, geminin, topoisomerase-2a, aurora kinase A, and phospho-histone H3 were rare or missing. These were likely silenced by upregulated CDK inhibitors p15INK4b, p16INK4a, p27KIP1, p53 through its effector p21WAF1 and possibly cyclin G1, consistent with the prevention of DNA replication. In conclusion, the upregulation of known and several novel cell cycle progression markers detected here clearly verify early replication activities in GCs, which are controlled by cell cycle arresting CDK inhibitors at G1 phase, and support the functional maturation of GCs in GCTB.

Number of Retrieval Attempts Rather Than Procedure Time Is Associated With Risk of Symptomatic Intracranial Hemorrhage.

Endovascular therapy is the standard of care in the treatment of acute ischemic stroke due to large-vessel occlusion. A direct association between the number of device passes and the occurrence of symptomatic intracranial hemorrhage (SICH) has been suggested. This study represents an in-depth investigation of the hypothesis that >3 retrieval attempts is associated with an increased rate of SICH in a large multicenter patient cohort. Two thousand six hundred eleven patients from the prospective German Stroke Registry were analyzed. Patients who received Endovascular therapy for acute large-vessel occlusion of the anterior circulation with known admission National Institutes of Health Stroke Scale and Alberta Stroke Program Early CT Score, final Thrombolysis in Cerebral Infarction, and number of retrieval passes were included. The primary outcome was defined as SICH. The secondary outcome was any type of radiologically confirmed intracranial hemorrhage within the first 24 hours. Multivariate mixed-effects models were used to adjust for cluster effects of the participating centers, as well as for confounders. Five hundred ninety-three patients fulfilled the inclusion criteria. The median number of retrieval passes was 2 [interquartile range, 1­3]. SICH occurred in 26 cases (4.4%), whereas intracranial hemorrhage was identified by neuroimaging in 85 (14.3%) cases. More than 3 retrieval passes was the strongest predictor for SICH (odds ratio, 3.61 [95% CI, 1.38­9.42], P=0.0089) following adjustment for age, admission National Institutes of Health Stroke Scale, admission Alberta Stroke Program Early CT Score, and Thrombolysis in Cerebral Infarction, as well as time from symptom onset to flow restoration. Baseline Alberta Stroke Program Early CT Score of 8 to 9 (odds ratio, 0.26 [95% CI, 0.07­0.89], P=0.032) or 10 (odds ratio, 0.21 [95% CI, 0.06­0.78], P=0.020) were significant protective factors against the occurrence of SICH. More than 3 retrieval attempts is associated with a significant increase in SICH risk, regardless of patient age, baseline National Institutes of Health Stroke Scale, or procedure time. This should be considered when deciding whether to continue a procedure, especially in patients with large baseline infarctions. URL: https://www.clinicaltrials.gov; Unique identifier: NCT03356392.

Comparative analysis of machine learning algorithms for computer-assisted reporting based on fully automated cross-lingual RadLex mappings.

Computer-assisted reporting (CAR) tools were suggested to improve radiology report quality by context-sensitively recommending key imaging biomarkers. However, studies evaluating machine learning (ML) algorithms on cross-lingual ontological (RadLex) mappings for developing embedded CAR algorithms are lacking. Therefore, we compared ML algorithms developed on human expert-annotated features against those developed on fully automated cross-lingual (German to English) RadLex mappings using 206 CT reports of suspected stroke. Target label was whether the Alberta Stroke Programme Early CT Score (ASPECTS) should have been provided (yes/no:154/52). We focused on probabilistic outputs of ML-algorithms including tree-based methods, elastic net, support vector machines (SVMs) and fastText (linear classifier), which were evaluated in the same 5 × fivefold nested cross-validation framework. This allowed for model stacking and classifier rankings. Performance was evaluated using calibration metrics (AUC, brier score, log loss) and -plots. Contextual ML-based assistance recommending ASPECTS was feasible. SVMs showed the highest accuracies both on human-extracted- (87%) and RadLex features (findings:82.5%; impressions:85.4%). FastText achieved the highest accuracy (89.3%) and AUC (92%) on impressions. Boosted trees fitted on findings had the best calibration profile. Our approach provides guidance for choosing ML classifiers for CAR tools in fully automated and language-agnostic fashion using bag-of-RadLex terms on limited expert-labelled training data.

Multiple breath washout (MBW) testing using sulfur hexafluoride: reference values and influence of anthropometric parameters.

BACKGROUND: Multiple breath washout (MBW) using sulfur hexafluoride (SF6) has the potential to reveal ventilation heterogeneity which is frequent in patients with obstructive lung disease and associated small airway dysfunction. However, reference data are scarce for this technique and mostly restricted to younger cohorts. We therefore set out to evaluate the influence of anthropometric parameters on SF6-MBW reference values in pulmonary healthy adults. METHODS: We evaluated cross-sectional data from 100 pulmonary healthy never-smokers and smokers (mean 51 (SD 20), range 20-88 years). Lung clearance index (LCI), acinar (Sacin) and conductive (Scond) ventilation heterogeneity were derived from triplicate SF6-MBW measurements. Global ventilation heterogeneity was calculated for the 2.5% (LCI2.5) and 5% (LCI5) stopping points. Upper limit of normal (ULN) was defined as the 95th percentile. RESULTS: Age was the only meaningful parameter influencing SF6-MBW parameters, explaining 47% (CI 33% to 59%) of the variance in LCI, 32% (CI 18% to 47%) in Sacin and 10% (CI 2% to 22%) in Scond. Mean LCI increases from 6.3 (ULN 7.4) to 8.8 (ULN 9.9) in subjects between 20 and 90 years. Smoking accounted for 2% (CI 0% to 8%) of the variability in LCI, 4% (CI 0% to 13%) in Sacin and 3% (CI 0% to 13%) in Scond. CONCLUSION: SF6-MBW outcome parameters showed an age-dependent increase from early adulthood to old age. The effect was most pronounced for global and acinar ventilation heterogeneity and smaller for conductive ventilation heterogeneity. No influence of height, weight and sex was seen. Reference values can now be provided for all important SF6-MBW outcome parameters over the whole age range. TRIAL REGISTRATION NUMBER: NCT04099225.

Good Clinical Outcome Decreases With Number of Retrieval Attempts in Stroke Thrombectomy: Beyond the First-Pass Effect.

BACKGROUND AND PURPOSE: Endovascular therapy is the standard of care in the treatment of acute ischemic stroke due to large-vessel occlusion. Often, more than one retrieval attempt is needed to achieve reperfusion. We aimed to quantify the influence of endovascular therapy on clinical outcome depending on the number of retrievals needed for successful reperfusion in a large multi-center cohort. METHODS: For this observational cohort study, 2611 patients from the prospective German Stroke Registry included between June 2015 and April 2018 were analyzed. Patients who received endovascular therapy for acute anterior circulation stroke with known admission National Institutes of Health Stroke Scale score and Alberta Stroke Program Early CT Score, final Thrombolysis in Cerebral Infarction score, and number of retrievals were included. Successful reperfusion was defined as a Thrombolysis in Cerebral Infarction score of 2b or 3. The primary outcome was defined as functional independence (modified Rankin Scale score of 0-2) at day 90. Multivariate mixed-effects models were used to adjust for cluster effects of the participating centers and confounders. RESULTS: The inclusion criteria were met by 1225 patients. The odds of good clinical outcome decreased with every retrieval attempt required for successful reperfusion: the first retrieval had the highest odds of good clinical outcome (adjusted odds ratio, 6.45 [95% CI, 4.0-10.4]), followed by the second attempt (adjusted odds ratio, 4.56 [95% CI, 2.7-7.7]), and finally the third (adjusted odds ratio, 3.16 [95% CI, 1.8-5.6]). CONCLUSIONS: Successful reperfusion within the first 3 retrieval attempts is associated with improved clinical outcome compared with patients without reperfusion. We conclude that at least 3 retrieval attempts should be performed in endovascular therapy of anterior circulation strokes. Registration: URL: https://www.clinicaltrials.gov. Unique identifier: NCT03356392.

Prospective Feasibility Study of Single-Shot Antibiotic Prophylaxis in Transrectal Focal Ablation of Prostate Cancer.

INTRODUCTION: Urinary tract infections (UTI) represent the most frequent complications after transrectal focal ablation of prostate cancer. Single-shot antibiotic prophylaxis for prevention has not yet been described. METHODS: In this cohort study of patients who received a high-intensity focused ultrasound (HIFU) ablation of prostate cancer within a registered prospective single-arm trial, we analyzed posttreatment UTI (≤30 days after HIFU) related to perioperative antibiotic management in an exploratory analysis: single-shot prophylaxis or targeted treatment for bacteriuria. Potential risk factors associated with UTI were evaluated by uni- and multivariate regression analyses. RESULTS: In total, 55 patients were eligible for analysis. Of these, 76.4% received antibiotic single-shot prophylaxis. UTI occurred in 10.7% of all patients, 5.4% developed fever, 3.6% required hospitalization. An antibiotic single-shot prophylaxis helped to protect 90.5% of men from infectious complications. Estimated effects indicate that a longer posttreatment catheterization (OR 3.38, 95% CI 0.47-27.08) and larger ablation volume (OR 4.85, 95% CI 0.61-107.49) might be associated with the highest risk for UTI after treatment. CONCLUSION: Single-shot antibiotic prophylaxis compared to a targeted antibiotic treatment showed a similar effectivity to prevent patients from infectious complications and should be considered as an element of antibiotic stewardship. Further research on risk factors and antibiotic strategies is required.