Automated Large Artery Occlusion Detection in Stroke: A Single-Center Validation Study of an Artificial Intelligence Algorithm

2021 ◽  
pp. 1-6
Author(s):  
Gabriel Rodrigues ◽  
Clara M. Barreira ◽  
Mehdi Bouslama ◽  
Diogo C. Haussen ◽  
Alhamza Al-Bayati ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Secondary Analysis ◽  
Large Artery ◽  
Single Center ◽  
Characteristic Analysis ◽  
Cohen’S Kappa ◽  
Accurate Identification ◽  
Vessel Occlusion ◽  
Test Characteristic ◽  
Cohen's Kappa

<b><i>Introduction:</i></b> Expediting notification of lesions in acute ischemic stroke (AIS) is critical. Limited availability of experts to assess such lesions and delays in large vessel occlusion (LVO) recognition can negatively affect outcomes. Artificial intelligence (AI) may aid LVO recognition and treatment. This study aims to evaluate the performance of an AI-based algorithm for LVO detection in AIS. <b><i>Methods:</i></b> Retrospective analysis of a database of AIS patients admitted in a single center between 2014 and 2019. Vascular neurologists graded computed tomography angiographies (CTAs) for presence and site of LVO. Studies were analyzed by the Viz-LVO Algorithm® version 1.4 – neural network programmed to detect occlusions from the internal carotid artery terminus (ICA-T) to the Sylvian fissure. Comparisons between human versus AI-based readings were done by test characteristic analysis and Cohen’s kappa. Primary analysis included ICA-T and/or middle cerebral artery (MCA)-M1 LVOs versus non-LVOs/more distal occlusions. Secondary analysis included MCA-M2 occlusions. <b><i>Results:</i></b> 610 CTAs were analyzed. The AI algorithm rejected 2.5% of the CTAs due to poor quality, which were excluded from the analysis. Viz-LVO identified ICA-T and MCA-M1 LVOs with a sensitivity of 87.6%, specificity of 88.5%, and accuracy of 87.9% (AUC 0.88, 95% CI: 0.85–0.92, <i>p</i> &#x3c; 0.001). Cohen’s kappa was 0.74. In the secondary analysis, the algorithm yielded a sensitivity of 80.3%, specificity of 88.5%, and accuracy of 82.7%. The mean run time of the algorithm was 2.78 ± 0.5 min. <b><i>Conclusion:</i></b> Automated AI reading allows for fast and accurate identification of LVO strokes with timely notification to emergency teams, enabling quick decision-making for reperfusion therapies or transfer to specialized centers if needed.

Abstract WP61: Artificial Intelligence for Automated Detection of Large Vessel Occlusion Strokes

Stroke ◽  
2020 ◽  
Vol 51 (Suppl_1) ◽  
Author(s):  
Catarina Perry da Câmara ◽  
Gabriel M Rodrigues ◽  
Clara Barreira ◽  
Mehdi Bouslama ◽  
Leonardo Pisani ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Detrimental Effect ◽  
Poor Quality ◽  
Automated Detection ◽  
Large Vessel ◽  
Large Vessel Occlusion ◽  
Accurate Identification ◽  
Vessel Occlusion ◽  
The Mean ◽  
Stroke Center

Introduction: Identification of Large Vessel Occlusion (LVO) in acute ischemic stroke (AIS) patients is critical for proper decision-making. Limited availability of trained experts and delays in LVO recognition can have a detrimental effect on outcomes. We sought to evaluate an artificial intelligence-based algorithm for LVO detection in AIS. Methods: A retrospective analysis of a prospectively-collected database of AIS patients admitted to a large volume stroke center between 2014-2018 was performed. Experienced vascular neurologists graded CTA for presence and site of LVO. Concurrently, studies were analyzed by the Viz-LVO Algorithm® version 1.4 (GA) - a convolutional neural network programmed to detect occlusions from the internal carotid artery terminus (ICA-T) to the sylvian fissure, which would include all MCA M1-segment and most M2-segment lesions. CTA readings were categorized as LVOs (ICA-T, MCA-M1, MCA-M2) versus non-LVOs/more distal occlusions. Comparisons between human and AI-based readings were done by accuracy analysis and calculating Cohen’s kappa. Results: A total of 610 CTAs were analyzed. The AI algorithm rejected 3.4% of the CTAs due to poor quality. Viz-LVO identified LVOs with an overall sensitivity of 81.3%, specificity of 87.8%, and accuracy of 83.2% (AUC 0.845 (95%CI:0.81-0.88, p<0.001). Table 1 shows the results per occlusion site. Accuracy was higher for ICA-T and M1 occlusions as compared to M2 occlusions. The mean run time of the algorithm was 2.78±0.5minutes. Conclusion: Our study demonstrates that automated AI reading allows for fast and accurate identification of LVO strokes. Future efforts should be made to improve the detection of the more distal occlusions.

scholarly journals Automated detection of lung nodules and coronary artery calcium using artificial intelligence on low-dose CT scans for lung cancer screening: accuracy and prognostic value

BMC Medicine ◽  
2021 ◽  
Vol 19 (1) ◽  
Author(s):  
Jordan Chamberlin ◽  
Madison R. Kocher ◽  
Jeffrey Waltz ◽  
Madalyn Snoddy ◽  
Natalie F. C. Stringer ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Lung Cancer ◽  
Coronary Artery ◽  
Coronary Artery Calcium ◽  
Low Dose ◽  
Chest Ct ◽  
Lung Nodules ◽  
Cohen’S Kappa ◽  
Cohen's Kappa ◽  
Cardiopulmonary Outcomes

Abstract Background Artificial intelligence (AI) in diagnostic radiology is undergoing rapid development. Its potential utility to improve diagnostic performance for cardiopulmonary events is widely recognized, but the accuracy and precision have yet to be demonstrated in the context of current screening modalities. Here, we present findings on the performance of an AI convolutional neural network (CNN) prototype (AI-RAD Companion, Siemens Healthineers) that automatically detects pulmonary nodules and quantifies coronary artery calcium volume (CACV) on low-dose chest CT (LDCT), and compare results to expert radiologists. We also correlate AI findings with adverse cardiopulmonary outcomes in a retrospective cohort of 117 patients who underwent LDCT. Methods A total of 117 patients were enrolled in this study. Two CNNs were used to identify lung nodules and CACV on LDCT scans. All subjects were used for lung nodule analysis, and 96 subjects met the criteria for coronary artery calcium volume analysis. Interobserver concordance was measured using ICC and Cohen’s kappa. Multivariate logistic regression and partial least squares regression were used for outcomes analysis. Results Agreement of the AI findings with experts was excellent (CACV ICC = 0.904, lung nodules Cohen’s kappa = 0.846) with high sensitivity and specificity (CACV: sensitivity = .929, specificity = .960; lung nodules: sensitivity = 1, specificity = 0.708). The AI findings improved the prediction of major cardiopulmonary outcomes at 1-year follow-up including major adverse cardiac events and lung cancer (AUCMACE = 0.911, AUCLung Cancer = 0.942). Conclusion We conclude the AI prototype rapidly and accurately identifies significant risk factors for cardiopulmonary disease on standard screening low-dose chest CT. This information can be used to improve diagnostic ability, facilitate intervention, improve morbidity and mortality, and decrease healthcare costs. There is also potential application in countries with limited numbers of cardiothoracic radiologists.

Abstract WP439: Who Benefits From CT-Angiography Previous Intravenous Thrombolysis?

Stroke ◽  
2016 ◽  
Vol 47 (suppl_1) ◽  
Author(s):  
Pedro Cardona ◽  
Helena Quesada ◽  
Blanca Lara ◽  
Nuria Cayuela ◽  
Paloma Mora ◽  
...  
Keyword(s):  
Endovascular Treatment ◽  
Arterial Occlusion ◽  
Intravenous Thrombolysis ◽  
Artery Occlusion ◽  
Clinical Group ◽  
Large Artery ◽  
Characteristic Analysis ◽  
Vessel Occlusion ◽  
Significant Difference ◽  
Distal Migration

Background: Endovascular treatment (EVT) is an effective treatment in strokes with persistent large artery occlusion despite previous intravenous thrombolisis (IVT) as rescue treatment. Performing computer tomography angiography (CTA) before IVT could allow early activation of neurointerventional teams; however routine CTA could delay unnecessary door-to-needle time of IVT and may be infeasible. Methods: We reviewed stroke code activations between May 2011 and June 2015 in our comprehensive stroke center and divided into groups based on NIHSS and patency of arterial occlusion according to non-enhanced CT on admission (dense artery sign or dot sign) and baseline CTA. We assessed patients treated with IVT and selected to EVT according to results in CTA post-IVT. We analyze percentage of recanalization or migration of thrombus after IVT alone and variables associated to successful treatment. Results: Of 2856 stroke codes registered during the study period 1810 were diagnosis of ischemic strokes. We treated 520 patients with IVT, 202 had a radiological evidence of large artery occlusion (55%M1, 32% M2, 5%TICA, 5%ICA, 3% basilar). Thirty-two percent of patients showed changes in CTA carried out after IVT(17% successfully recanalized, 15% distal migration of thrombus) so they were not selected to endovascular treatment. There were significant difference between M1 and M2 occlusion regarding changes in CTA after IVT (23% vs 70%; p<0.001). In multivariate logistic regression a baseline score NIH<10 was associated with higher percentage of recanalization with rtPA despite signs of large vessel occlusion (78% vs 32%; p:0.001). In receiver operating characteristic analysis higher baseline NIH was associated with persistent occlusion after IVT (area under curve=0.79;95% CI, 0.6-0.9; P:0.001) with optima threshold of 10 ( Sensivity 84%, Specificity 74%). Conclusions: We consider defer CTA angiography until after IVT in stroke code patients with moderate clinical impairment (NIH<10) or M2-segment occlusion, because they achieve a high percentage of arterial recanalization. CTA previous IVT could be unnecessary, provide unreliable information and delay IVT in that clinical group but could be useful to plan EVT in patients with higher NIH scores.

Real-World Experience with Artificial Intelligence-Based Triage in Transferred Large Vessel Occlusion Stroke Patients

2021 ◽  
pp. 1-6
Author(s):  
Jacob R. Morey ◽  
Xiangnan Zhang ◽  
Kurt A. Yaeger ◽  
Emily Fiano ◽  
Naoum Fares Marayati ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Clinical Outcomes ◽  
Warning System ◽  
Large Vessel ◽  
Time Interval ◽  
Large Vessel Occlusion ◽  
Time Intervals ◽  
Vessel Occlusion ◽  
Left Behind ◽  
Stroke Center

<b><i>Background and Purpose:</i></b> Randomized controlled trials have demonstrated the importance of time to endovascular therapy (EVT) in clinical outcomes in large vessel occlusion (LVO) acute ischemic stroke. Delays to treatment are particularly prevalent when patients require a transfer from hospitals without EVT capability onsite. A computer-aided triage system, Viz LVO, has the potential to streamline workflows. This platform includes an image viewer, a communication system, and an artificial intelligence (AI) algorithm that automatically identifies suspected LVO strokes on CTA imaging and rapidly triggers alerts. We hypothesize that the Viz application will decrease time-to-treatment, leading to improved clinical outcomes. <b><i>Methods:</i></b> A retrospective analysis of a prospectively maintained database was assessed for patients who presented to a stroke center currently utilizing Viz LVO and underwent EVT following transfer for LVO stroke between July 2018 and March 2020. Time intervals and clinical outcomes were compared for 55 patients divided into pre- and post-Viz cohorts. <b><i>Results:</i></b> The median initial door-to-neuroendovascular team (NT) notification time interval was significantly faster (25.0 min [IQR = 12.0] vs. 40.0 min [IQR = 61.0]; <i>p</i> = 0.01) with less variation (<i>p</i> &#x3c; 0.05) following Viz LVO implementation. The median initial door-to-skin puncture time interval was 25 min shorter in the post-Viz cohort, although this was not statistically significant (<i>p</i> = 0.15). <b><i>Conclusions:</i></b> Preliminary results have shown that Viz LVO implementation is associated with earlier, more consistent NT notification times. This application can serve as an early warning system and a failsafe to ensure that no LVO is left behind.

scholarly journals An update on image forming methods: structure analysis and Gestalt evaluation of images from rocket lettuce with shading, N supply, organic or mineral fertilization, and biodynamic preparations

2021 ◽  
Author(s):  
Miriam Athmann ◽  
Roya Bornhütter ◽  
Nicolaas Busscher ◽  
Paul Doesburg ◽  
Uwe Geier ◽  
...  
Keyword(s):  
Copper Chloride ◽  
Organic Fertilizer ◽  
Structural Features ◽  
Fertilizer Treatment ◽  
Friedman Test ◽  
Mineral Fertilization ◽  
Cohen’S Kappa ◽  
N Supply ◽  
Cohen's Kappa ◽  
Forming Methods

AbstractIn the image forming methods, copper chloride crystallization (CCCryst), capillary dynamolysis (CapDyn), and circular chromatography (CChrom), characteristic patterns emerge in response to different food extracts. These patterns reflect the resistance to decomposition as an aspect of resilience and are therefore used in product quality assessment complementary to chemical analyses. In the presented study, rocket lettuce from a field trial with different radiation intensities, nitrogen supply, biodynamic, organic and mineral fertilization, and with or without horn silica application was investigated with all three image forming methods. The main objective was to compare two different evaluation approaches, differing in the type of image forming method leading the evaluation, the amount of factors analyzed, and the deployed perceptual strategy: Firstly, image evaluation of samples from all four experimental factors simultaneously by two individual evaluators was based mainly on analyzing structural features in CapDyn (analytical perception). Secondly, a panel of eight evaluators applied a Gestalt evaluation imbued with a kinesthetic engagement of CCCryst patterns from either fertilization treatments or horn silica treatments, followed by a confirmatory analysis of individual structural features. With the analytical approach, samples from different radiation intensities and N supply levels were identified correctly in two out of two sample sets with groups of five samples per treatment each (Cohen’s kappa, p = 0.0079), and the two organic fertilizer treatments were differentiated from the mineral fertilizer treatment in eight out of eight sample sets with groups of three manure and two minerally fertilized samples each (Cohen’s kappa, p = 0.0048). With the panel approach based on Gestalt evaluation, biodynamic fertilization was differentiated from organic and mineral fertilization in two out of two exams with 16 comparisons each (Friedman test, p < 0.001), and samples with horn silica application were successfully identified in two out of two exams with 32 comparisons each (Friedman test, p < 0.001). Further research will show which properties of the food decisive for resistance to decomposition are reflected by analytical and Gestalt criteria, respectively, in CCCryst and CapDyn images.

scholarly journals Endovascular therapy in patients with large vessel occlusion due to cardioembolism versus large-artery atherosclerosis

2021 ◽  
Vol 14 ◽  
pp. 175628642199901
Author(s):  
Meredeth Zotter ◽  
Eike I. Piechowiak ◽  
Rupashani Balasubramaniam ◽  
Rascha Von Martial ◽  
Kotryna Genceviciute ◽  
...  
Keyword(s):  
Heart Failure ◽  
Endovascular Therapy ◽  
Independent Predictor ◽  
Intravenous Thrombolysis ◽  
Stent Retriever ◽  
Large Vessel ◽  
Large Artery ◽  
Large Vessel Occlusion ◽  
Vessel Occlusion ◽  
Significant Difference

Background and aims: To investigate whether stroke aetiology affects outcome in patients with acute ischaemic stroke who undergo endovascular therapy. Methods: We retrospectively analysed patients from the Bernese Stroke Centre Registry (January 2010–September 2018), with acute large vessel occlusion in the anterior circulation due to cardioembolism or large-artery atherosclerosis, treated with endovascular therapy (±intravenous thrombolysis). Results: The study included 850 patients (median age 77.4 years, 49.3% female, 80.1% with cardioembolism). Compared with those with large-artery atherosclerosis, patients with cardioembolism were older, more often female, and more likely to have a history of hypercholesterolaemia, atrial fibrillation, current smoking (each p < 0.0001) and higher median National Institutes of Health Stroke Scale (NIHSS) scores on admission ( p = 0.030). They were more frequently treated with stent retrievers ( p = 0.007), but the median number of stent retriever attempts was lower ( p = 0.016) and fewer had permanent stent placements ( p ⩽ 0.004). Univariable analysis showed that patients with cardioembolism had worse 3-month survival [72.7% versus 84%, odds ratio (OR) = 0.51; p = 0.004] and modified Rankin scale (mRS) score shift ( p = 0.043) and higher rates of post-interventional heart failure (33.5% versus 18.5%, OR = 2.22; p < 0.0001), but better modified thrombolysis in cerebral infarction (mTICI) score shift ( p = 0.025). Excellent (mRS = 0–1) 3-month outcome, successful reperfusion (mTICI = 2b–3), symptomatic intracranial haemorrhage and Updated Charlson Comorbidity Index were similar between groups. Propensity-matched analysis found no statistically significant difference in outcome between stroke aetiology groups. Stroke aetiology was not an independent predictor of favourable mRS score shift, but lower admission NIHSS score, younger age and independence pre-stroke were (each p < 0.0001). Stroke aetiology was not an independent predictor of heart failure, but older age, admission antithrombotics and dependence pre-stroke were (each ⩽0.027). Stroke aetiology was not an independent predictor of favourable mTICI score shift, but application of stent retriever and no permanent intracranial stent placement were (each ⩽0.044). Conclusion: We suggest prospective studies to further elucidate differences in reperfusion and outcome between patients with cardioembolism and large-artery atherosclerosis.

scholarly journals Prediction of Streptococcus uberis clinical mastitis treatment success in dairy herds by means of mass spectrometry and machine-learning

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Alexandre Maciel-Guerra ◽  
Necati Esener ◽  
Katharina Giebel ◽  
Daniel Lea ◽  
Martin J. Green ◽  
...  
Keyword(s):  
Machine Learning ◽  
Mass Spectrometry ◽  
Treatment Success ◽  
Clinical Mastitis ◽  
Supervised Machine Learning ◽  
Structural Protein ◽  
Cohen’S Kappa ◽  
Maldi Tof ◽  
Streptococcus Uberis ◽  
Cohen's Kappa

AbstractStreptococcus uberis is one of the leading pathogens causing mastitis worldwide. Identification of S. uberis strains that fail to respond to treatment with antibiotics is essential for better decision making and treatment selection. We demonstrate that the combination of supervised machine learning and matrix-assisted laser desorption ionization/time of flight (MALDI-TOF) mass spectrometry can discriminate strains of S. uberis causing clinical mastitis that are likely to be responsive or unresponsive to treatment. Diagnostics prediction systems trained on 90 individuals from 26 different farms achieved up to 86.2% and 71.5% in terms of accuracy and Cohen’s kappa. The performance was further increased by adding metadata (parity, somatic cell count of previous lactation and count of positive mastitis cases) to encoded MALDI-TOF spectra, which increased accuracy and Cohen’s kappa to 92.2% and 84.1% respectively. A computational framework integrating protein–protein networks and structural protein information to the machine learning results unveiled the molecular determinants underlying the responsive and unresponsive phenotypes.

scholarly journals Endovascular thrombectomy time metrics in the era of COVID-19: observations from the Society of Vascular and Interventional Neurology Multicenter Collaboration

2021 ◽  
pp. neurintsurg-2020-017205
Author(s):  
Alexandra L Czap ◽  
Alicia M Zha ◽  
Jacob Sebaugh ◽  
Ameer E Hassan ◽  
Julie G Shulman ◽  
...  
Keyword(s):  
Care Delivery ◽  
Control Period ◽  
Secondary Analysis ◽  
Primary Analysis ◽  
Vessel Occlusion ◽  
Endovascular Thrombectomy ◽  
Final Multivariable Model ◽  
Multivariable Regression ◽  
Multi Center Study ◽  
Multivariable Adjustment

BackgroundUnprecedented workflow shifts during the coronavirus disease 2019 (COVID-19) pandemic have contributed to delays in acute care delivery, but whether it adversely affected endovascular thrombectomy metrics in acute large vessel occlusion (LVO) is unknown.MethodsWe performed a retrospective review of observational data from 14 comprehensive stroke centers in nine US states with acute LVO. EVT metrics were compared between March to July 2019 against March to July 2020 (primary analysis), and between state-specific pre-peak and peak COVID-19 months (secondary analysis), with multivariable adjustment.ResultsOf the 1364 patients included in the primary analysis (51% female, median NIHSS 14 [IQR 7–21], and 74% of whom underwent EVT), there was no difference in the primary outcome of door-to-puncture (DTP) time between the 2019 control period and the COVID-19 period (median 71 vs 67 min, P=0.10). After adjustment for variables associated with faster DTP, and clustering by site, there remained a trend toward shorter DTP during the pandemic (βadj=-73.2, 95% CI −153.8–7.4, Pp=0.07). There was no difference in DTP times according to local COVID-19 peaks vs pre-peak months in unadjusted or adjusted multivariable regression (βadj=-3.85, 95% CI −36.9–29.2, P=0.80). In this final multivariable model (secondary analysis), faster DTP times were significantly associated with transfer from an outside institution (βadj=-46.44, 95% CI −62.8 to – -30.0, P<0.01) and higher NIHSS (βadj=-2.15, 95% CI −4.2to – -0.1, P=0.05).ConclusionsIn this multi-center study, there was no delay in EVT among patients treated for intracranial occlusion during the COVID-19 era compared with the pre-COVID era.

scholarly journals The accuracy of initial diagnoses in coma: an observational study in 835 patients with non-traumatic disorder of consciousness

2021 ◽  
Vol 29 (1) ◽  
Author(s):  
Maximilian Lutz ◽  
Martin Möckel ◽  
Tobias Lindner ◽  
Christoph J. Ploner ◽  
Mischa Braun ◽  
...  
Keyword(s):  
Emergency Care ◽  
Final Diagnosis ◽  
Class I ◽  
Prehospital Emergency Care ◽  
Unknown Etiology ◽  
Care Providers ◽  
Cohen’S Kappa ◽  
Disorder Of Consciousness ◽  
Cohen's Kappa ◽  
Suspected Diagnosis

Abstract Background Management of patients with coma of unknown etiology (CUE) is a major challenge in most emergency departments (EDs). CUE is associated with a high mortality and a wide variety of pathologies that require differential therapies. A suspected diagnosis issued by pre-hospital emergency care providers often drives the first approach to these patients. We aim to determine the accuracy and value of the initial diagnostic hypothesis in patients with CUE. Methods Consecutive ED patients presenting with CUE were prospectively enrolled. We obtained the suspected diagnoses or working hypotheses from standardized reports given by prehospital emergency care providers, both paramedics and emergency physicians. Suspected and final diagnoses were classified into I) acute primary brain lesions, II) primary brain pathologies without acute lesions and III) pathologies that affected the brain secondarily. We compared suspected and final diagnosis with percent agreement and Cohen’s Kappa including sub-group analyses for paramedics and physicians. Furthermore, we tested the value of suspected and final diagnoses as predictors for mortality with binary logistic regression models. Results Overall, suspected and final diagnoses matched in 62% of 835 enrolled patients. Cohen’s Kappa showed a value of κ = .415 (95% CI .361–.469, p < .005). There was no relevant difference in diagnostic accuracy between paramedics and physicians. Suspected diagnoses did not significantly interact with in-hospital mortality (e.g., suspected class I: OR .982, 95% CI .518–1.836) while final diagnoses interacted strongly (e.g., final class I: OR 5.425, 95% CI 3.409–8.633). Conclusion In cases of CUE, the suspected diagnosis is unreliable, regardless of different pre-hospital care providers’ qualifications. It is not an appropriate decision-making tool as it neither sufficiently predicts the final diagnosis nor detects the especially critical comatose patient. To avoid the risk of mistriage and unnecessarily delayed therapy, we advocate for a standardized diagnostic work-up for all CUE patients that should be triggered by the emergency symptom alone and not by any suspected diagnosis.

Sign in / Sign up

Export Citation Format

Share Document