Ischemic Stroke Subtype Classification: An Asian Viewpoint

Background and Purpose— In Japan, nearly half of ischemic stroke patients receive edaravone for acute treatment. The purpose of this study was to assess the effect of edaravone on neurological symptoms in patients with ischemic stroke stratified by stroke subtype. Methods— Study subjects were 61 048 patients aged 18 years or older who were hospitalized ≤14 days after onset of an acute ischemic stroke and were registered in the Japan Stroke Data Bank, a hospital-based multicenter stroke registration database, between June 2001 and July 2013. Patients were stratified according to ischemic stroke subtype (large-artery atherosclerosis, cardioembolism, small-vessel occlusion, and cryptogenic/undetermined) and then divided into 2 groups (edaravone-treated and no edaravone). Neurological symptoms were evaluated using the National Institutes of Health Stroke Scale (NIHSS). The primary outcome was changed in neurological symptoms during the hospital stay (ΔNIHSS=NIHSS score at discharge−NIHSS score at admission). Data were analyzed using multivariate linear regression with inverse probability of treatment weighting after adjusting for the following confounding factors: age, gender, and systolic and diastolic blood pressure at the start of treatment, NIHSS score at admission, time from stroke onset to hospital admission, infarct size, comorbidities, concomitant medication, clinical department, history of smoking, alcohol consumption, and history of stroke. Results— After adjusting for potential confounders, the improvement in NIHSS score from admission to discharge was greater in the edaravone-treated group than in the no edaravone group for all ischemic stroke subtypes (mean [95% CI] difference in ΔNIHSS: −0.46 [−0.75 to −0.16] for large-artery atherosclerosis, −0.64 [−1.09 to −0.2] for cardioembolism, and −0.25 [−0.4 to −0.09] for small-vessel occlusion). Conclusions— For any ischemic stroke subtype, edaravone use (compared with no use) was associated with a greater improvement in neurological symptoms, although the difference was small (<1 point NIHSS) and of limited clinical significance.

Download Full-text

Five-year readmission and mortality differ by ischemic stroke subtype

Journal of the Neurological Sciences ◽

10.1016/j.jns.2019.06.007 ◽

2019 ◽

Vol 403 ◽

pp. 31-37 ◽

Cited By ~ 1

Author(s):

Anna Therese Bjerkreim ◽

Andrej Netland Khanevski ◽

Lars Thomassen ◽

Henriette Aurora Selvik ◽

Ulrike Waje-Andreassen ◽

...

Keyword(s):

Ischemic Stroke ◽

Stroke Subtype ◽

Ischemic Stroke Subtype

Download Full-text

Interpretable CNN for ischemic stroke subtype classification with active model adaptation

BMC Medical Informatics and Decision Making ◽

10.1186/s12911-021-01721-5 ◽

2022 ◽

Vol 22 (1) ◽

Author(s):

Shuo Zhang ◽

Jing Wang ◽

Lulu Pei ◽

Kai Liu ◽

Yuan Gao ◽

...

Keyword(s):

Ischemic Stroke ◽

Selection Criterion ◽

Fine Tuning ◽

Data Types ◽

Stroke Subtype ◽

Subtype Classification ◽

Active Selection ◽

Sequential Experiments ◽

Evaluation Metric ◽

Toast Classification

Abstract Background TOAST subtype classification is important for diagnosis and research of ischemic stroke. Limited by experience of neurologist and time-consuming manual adjudication, it is a big challenge to finish TOAST classification effectively. We propose a novel active deep learning architecture to classify TOAST. Methods To simulate the diagnosis process of neurologists, we drop the valueless features by XGB algorithm and rank the remaining ones. Utilizing active learning framework, we propose a novel causal CNN, in which it combines with a mixed active selection criterion to optimize the uncertainty of samples adaptively. Meanwhile, KL-focal loss derived from the enhancement of Focal loss by KL regularization is introduced to accelerate the iterative fine-tuning of the model. Results To evaluate the proposed method, we construct a dataset which consists of totally 2310 patients. In a series of sequential experiments, we verify the effectiveness of each contribution by different evaluation metrics. Experimental results show that the proposed method achieves competitive results on each evaluation metric. In this task, the improvement of AUC is the most obvious, reaching 77.4. Conclusions We construct a backbone causal CNN to simulate the neurologist process of that could enhance the internal interpretability. The research on clinical data also indicates the potential application value of this model in stroke medicine. Future work we would consider various data types and more comprehensive patient types to achieve fully automated subtype classification.

Download Full-text