Convolutional Neural Network Model to Detect COVID-19 Patients Utilizing Chest X-ray Images

  1. Md. Shahriare Satu
  2. Khair Ahammed
  3. Mohammad Zoynul Abedin
  4. Md. Auhidur Rahman
  5. Sheikh Mohammed Shariful Islam
  6. AKM Azad
  7. Salem A. Alyami
  8. Mohammad Ali Moni

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

This study aims to propose a deep learning model to detect COVID-19 positive cases more precisely utilizing chest X-ray images. We have collected and merged all the publicly available chest X-ray datasets of COVID-19 infected patients from Kaggle and Github, and pre-processed it using random sampling approach. Then, we proposed and applied an enhanced convolutional neural network (CNN) model to this dataset and obtained a 94.03% accuracy, 95.52% AUC and 94.03% f-measure for detecting COVID-19 positive patients. We have also performed a comparative performance between our proposed CNN model with several state-of-the-art machine learning classifiers including support vector machine, random forest, k-nearest neighbor, logistic regression, gaussian naïve bayes, bernoulli naïve bayes, decision tree, Xgboost, multilayer perceptron, nearest centroid and perceptron as well as deep learning and pre-trained models such as deep neural network, residual neural network, visual geometry group network 16, and inception network V3 were employed, where our model yielded outperforming results compared to all other models. While evaluating the performance of our models, we have emphasized on specificity along with accuracy to identify non-COVID-19 individuals more accurately, which may potentially facilitate the early detection of COVID-19 patients for their preliminary screening, especially in under-resourced health infrastructure with insufficient PCR testing systems and testing facilities. Moreover, this model could also be applicable to the cases of other lung infections.

Article activity feed

  1. Version published to 10.1101/2020.06.07.20124594v2 on medRxiv
  2. ScreenIT

    SciScore for 10.1101/2020.06.07.20124594: (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

    Software and Algorithms
    SentencesResources
    Therefore, these classifiers were employed 10 fold cross validation procedure using python programming language.
    python
    suggested: (IPython, RRID:SCR_001658)

    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.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  3. Version published to 10.1101/2020.06.07.20124594v1 on medRxiv