COVID-19 Chest X-Ray Image Classification Using Deep Learning
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (ScreenIT)
Abstract
The rise of the coronavirus disease 2019 (COVID-19) pandemic has made it necessary to improve existing medical screening and clinical management of this disease. While COVID-19 patients are known to exhibit a variety of symptoms, the major symptoms include fever, cough, and fatigue. Since these symptoms also appear in pneumonia patients, this creates complications in COVID-19 detection especially during the flu season. Early studies identified abnormalities in chest X-ray images of COVID-19 infected patients that could be beneficial for disease diagnosis. Therefore, chest X-ray image-based disease classification has emerged as an alternative to aid medical diagnosis. However, manual detection of COVID-19 from a set of chest X-ray images comprising both COVID-19 and pneumonia cases is cumbersome and prone to human error. Thus, artificial intelligence techniques powered by deep learning algorithms, which learn from radiography images and predict presence of COVID-19 have potential to enhance current diagnosis process. Towards this purpose, here we implemented a set of deep learning pre-trained models such as ResNet, VGG, Inception and EfficientNet in conjunction with developing a computer vision AI system based on our own convolutional neural network (CNN) model: Deep Learning in Healthcare (DLH)-COVID. All these CNN models cater to image classification exercise. We used publicly available resources of 6,432 images and further strengthened our model by tuning hyperparameters to provide better generalization during the model validation phase. Our final DLH-COVID model yielded the highest accuracy of 96% in detection of COVID-19 from chest X-ray images when compared to images of both pneumonia-affected and healthy individuals. Given the practicality of acquiring chest X-ray images by patients, we also developed a web application (link: https://toad.li/xray ) based on our model to directly enable users to upload chest X-ray images and detect the presence of COVID-19 within a few seconds. Taken together, here we introduce a state-of-the-art artificial intelligence-based system for efficient COVID-19 detection and a user-friendly application that has the capacity to become a rapid COVID-19 diagnosis method in the near future.
Article activity feed
-
SciScore for 10.1101/2021.07.15.21260605: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
NIH rigor criteria are not applicable to paper type.Table 2: Resources
Experimental Models: Organisms/Strains Sentences Resources We used VGG16, VGG19 [17], ResNet18, ResNet34, ResNet50, RestNet101[16], Inception [18] and EfficientNet [19] that are aligned with the transfer learning method. VGG19suggested: NoneSoftware and Algorithms Sentences Resources (1) Learning rate scheduling method: We used PyTorch library module optim() that provides several methods to adjust learning rate based on the number of epochs during model training phase [31]. PyTorchsuggested: (PyTorch, RRID:SCR_018536)Results from OddPub: Thank you for sharing your code and data.
Results from LimitationRecognizer: An explicit section …SciScore for 10.1101/2021.07.15.21260605: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
NIH rigor criteria are not applicable to paper type.Table 2: Resources
Experimental Models: Organisms/Strains Sentences Resources We used VGG16, VGG19 [17], ResNet18, ResNet34, ResNet50, RestNet101[16], Inception [18] and EfficientNet [19] that are aligned with the transfer learning method. VGG19suggested: NoneSoftware and Algorithms Sentences Resources (1) Learning rate scheduling method: We used PyTorch library module optim() that provides several methods to adjust learning rate based on the number of epochs during model training phase [31]. PyTorchsuggested: (PyTorch, RRID:SCR_018536)Results from OddPub: Thank you for sharing your code and data.
Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.Results from TrialIdentifier: No clinical trial numbers were referenced.
Results from Barzooka: We did not find any issues relating to the usage of bar graphs.
Results from JetFighter: We did not find any issues relating to colormaps.
Results from rtransparent:- Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
- Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
- No protocol registration statement was detected.
Results from scite Reference Check: We found no unreliable references.
-
