Deep learning-based identification of acute ischemic core and deficit from non-contrast CT and CTA

Wang, Chengyan; Shi, Zhang; Yang, Ming; Huang, Libing; Fang, W.; Jiang, Li; Ding, Jing; Wang, He

doi:10.1177/0271678x211023660

Cited by 14 publications

(16 citation statements)

References 47 publications

(81 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Like other studies combing machine learning and disease models, 27,55 it is hoped the findings of this study would be valuable for clinical practice and to improve patient outcomes. A key point making future translational research promising is that the optimal algorithms and predictive models showed good and stable performance in other external datasets.…”

Section: Discussionmentioning

confidence: 91%

“…Four subsets were used as training data and the remaining one as the validation set. The cross‐validation process was repeated five times, and each of the five folds was used once as validation data 27,28 . The results were then averaged to produce a single estimation.…”

Section: Methodsmentioning

confidence: 99%

“…The cross‐validation process was repeated five times, and each of the five folds was used once as validation data. 27 , 28 The results were then averaged to produce a single estimation. To overcome overfitting due to group imbalance when predicting hypokalemia of different severity, a propensity score matching approach was used to balance the gender and age of patients in a 1:1 ratio.…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Prediction of in‐hospital hypokalemia using machine learning and first hospitalization day records in patients with traumatic brain injury

Zhou

Huang

et al. 2022

CNS Neurosci Ther

View full text Add to dashboard Cite

Aims Hypokalemia is a common complication following traumatic brain injury, which may complicate treatment and lead to unfavorable outcomes. Identifying patients at risk of hypokalemia on the first day of admission helps to implement prophylactic treatment, reduce complications, and improve prognosis. Methods This multicenter retrospective study was performed between January 2017 and December 2020 using the electronic medical records of patients admitted due to traumatic brain injury. A propensity score matching approach was adopted with a ratio of 1:1 to overcome overfitting and data imbalance during subgroup analyses. Five machine learning algorithms were applied to generate a best‐performed prediction model for in‐hospital hypokalemia. The internal fivefold cross‐validation and external validation were performed to demonstrate the interpretability and generalizability. Results A total of 4445 TBI patients were recruited for analysis and model generation. Hypokalemia occurred in 46.55% of recruited patients and the incidences of mild, moderate, and severe hypokalemia were 32.06%, 12.69%, and 1.80%, respectively. Hypokalemia was associated with increased mortality, while severe hypokalemia cast greater impacts. The logistic regression algorithm had the best performance in predicting decreased serum potassium and moderate‐to‐severe hypokalemia, with an AUC of 0.73 ± 0.011 and 0.74 ± 0.019, respectively. The prediction model was further verified using two external datasets, including our previous published data and the open‐assessed Medical Information Mart for Intensive Care database. Linearized calibration curves showed no statistical difference (p > 0.05) with perfect predictions. Conclusions The occurrence of hypokalemia following traumatic brain injury can be predicted by first hospitalization day records and machine learning algorithms. The logistic regression algorithm showed an optimal predicting performance verified by both internal and external validation.

show abstract

Section: Discussionmentioning

confidence: 91%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Prediction of in‐hospital hypokalemia using machine learning and first hospitalization day records in patients with traumatic brain injury

Zhou

Huang

et al. 2022

CNS Neurosci Ther

View full text Add to dashboard Cite

show abstract

“…Pearson correlation coefficients reported range from 0.44 to 0.76. 5,6,9,10,11 The Dice similarity coefficient reported range from 0.32 to 0.45. 6,10 Among the previous reports, the model reported by Qui et al 5 performed the best.…”

Section: Discussionmentioning

confidence: 99%

Automatic Ischemic Core Estimation Based on Noncontrast-Enhanced Computed Tomography

Nishi

Ishii

Tsuji

et al. 2023

Stroke

View full text Add to dashboard Cite

Background: Evaluating the extent of ischemic change is an important step in deciding whether to use thrombolysis or mechanical thrombectomy, but the current standard method, Alberta Stroke Program Early CT Score, is semiquantitative and has low consistency among raters. We aim to create and test a fully automated machine learning–based ischemic core segmentation model using only noncontrast-enhanced computed tomography images. Methods: In this multicenter retrospective study, patients with anterior circulation acute ischemic stroke who received both computed tomography (CT) and magnetic resonance imaging before thrombolysis or recanalization treatment between 2013 and 2019 were included. On CT, the ischemic core was manually delineated using the diffusion-weighted image and apparent diffusion coefficient maps. A deep learning–based ischemic core segmentation model (DL model) was developed using data from 3 institutions (n=272), and the model performance was validated using data from 3 institutions (n=106 Results: The median time ).between CT and magnetic resonance imaging in the validation cohort was 18 min. The DL model calculated ischemic core volume was significantly correlated with the reference standard (intraclass correlation coefficient, 0.90, P <0.01). Both the early time window (≤4.5 hours from onset; intraclass correlation coefficient, 0.90, P <0.01) and the late time window (>4.5 hours from onset; intraclass correlation coefficient, 0.93, P <0.01) had significant correlations. The median difference in ivolume between the model and the reference standard was 4.7 mL (interquartile range, 0.8–12.4 mL). The DL model performed well in distinguishing large ischemic cores (>70 mL), with a sensitivity of 84.2%, specificity of 97.7%, and area under the curve of 0.91. Conclusions: The deep learning–based ischemic core segmentation model, which was based on noncontrast-enhanced CT, demonstrated high accuracy in assessing ischemic core volume in patients with anterior circulation acute ischemic stroke.

show abstract

“…Contextual information's are used to identify Stroke signs [18,19]. Using widely available NCCT and CT angiography (CTA) data, deep learning can be employed to identify lesions [20].…”

Section: Introductionmentioning

confidence: 99%

An Ensemble Deep Learning Network in Classifying the Early CT Slices of Ischemic Stroke Patients

Rajendran¹,

Radhakrishnan²,

Viswanathan³

2022

View full text Add to dashboard Cite

The human brain is the body's most complicated organ. Constant blood flow is essential for the sustained functioning of the brain. A blocked blood vessel's interruption of blood supply prevents oxygen and nutrients to the brain tissues. This results in a life-threatening brain disease called Ischemic Stroke. Computed Tomography (CT) images are widely used in the diagnosis of Ischemic Stroke because of their faster acquisition and compatibility with most life support devices. CT acquired from the patients who arrived with stroke symptoms is Primary CT (PCT). After some hours CT taken for the same patient is Secondary CT (SCT). Stroke lesions may not be visible in PCT, whereas visible in SCT. Learning the features automatically using a Convolutional Neural Network (CNN) is essential to classify normal and abnormal CT slices. These networks are capable of learning the global features effectively for image classification. Though this CNN approach works, achieving desired accuracy was challenging. Different architectures considered for this CNN experimentation are VGG1, VGG2, VGG3, VGG16, InceptionV3, and ResNet50. This novel work provides a detailed explanation of the three experiments conducted using PCT and SCT slices. Three experiments are conducted using SCT and PCT slices. The pretrained VGG16, ResNet50, and InceptionV3 networks with the ImageNet database are applied as a first approach. Both SCT and PCT slices are used for testing alone. It resulted in 49.22%, 47.076% and 49.36% classification accuracy. In the second approach, different models were trained for classification from PCT and SCT slices. This includes the networks like VGG1, VGG2, VGG3, VGG16, VGG16 with dropout, ResNet, ResNet50 with lambda regularization, InceptionV3, and InceptionV3 with lambda regularization. The accuracies achieved are 68%, 69.4%, 72%, 78.2%, 79.1%, 77%, 77.8%, 79.6% and 80.1%. The accuracy was improved with dropout and lambda regularization. The networks with high accuracy are selected and an ensemble model is developed as a third approach. ResNet50, VGG16, and InceptionV3 are combined to form an ensemble network. This ensemble network yielded an accuracy of 81.98% when SCT and PCT slices are used for both training and testing. And produced 74% accuracy when PCT slices alone were used. Also produced 93.76% accuracy when SCT slices alone were used.

show abstract

Deep learning-based identification of acute ischemic core and deficit from non-contrast CT and CTA

Cited by 14 publications

References 47 publications

Prediction of in‐hospital hypokalemia using machine learning and first hospitalization day records in patients with traumatic brain injury

Prediction of in‐hospital hypokalemia using machine learning and first hospitalization day records in patients with traumatic brain injury

Automatic Ischemic Core Estimation Based on Noncontrast-Enhanced Computed Tomography

An Ensemble Deep Learning Network in Classifying the Early CT Slices of Ischemic Stroke Patients

Contact Info

Product

Resources

About