Related Experiment Video
Updated: Jul 30, 2025
04:17
DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning
Published on: May 10, 2024
829
A Real Time Method for Distinguishing COVID-19 Utilizing 2D-CNN and Transfer Learning
Abida Sultana1, Md Nahiduzzaman1,2, Sagor Chandro Bakchy1
1Department of Electrical & Computer Engineering, Rajshahi University of Engineering & Technology, Rajshahi 6204, Bangladesh.
Sensors (Basel, Switzerland)
|May 13, 2023
Summary
A novel deep learning model accurately identifies COVID-19 and other lung diseases from Chest X-rays (CXRs). This AI approach offers faster, more reliable detection than PCR tests for improved patient care and epidemic control.
Area of Science:
- Medical Imaging and Artificial Intelligence
- Radiology and Diagnostic Imaging
- Computational Biology and Bioinformatics
Background:
- The Polymerase Chain Reaction (PCR) test for COVID-19 is time-consuming and requires expertise, limiting its widespread use for rapid diagnosis.
- Chest X-ray (CXR) imaging offers a real-time alternative for detecting COVID-19 and other lung infections.
- Deep learning (DL) models show significant promise for automated disease classification from medical images.
Purpose of the Study:
- To develop and evaluate deep learning models for efficient identification of COVID-19 and other lung disorders using CXR images.
- To compare the performance of a proposed Convolutional Neural Network (CNN) model against established DL architectures (VGG-16, VGG-19, Inception-v3).
Main Methods:
- A dataset of 18,564 CXR images across seven categories (healthy, fibrosis, lung opacity, viral pneumonia, bacterial pneumonia, COVID-19, tuberculosis) was curated.
- Four DL models, including a proposed CNN, were trained and evaluated for disease classification.
- Performance metrics included accuracy, precision, recall, F1-score, Area Under the Curve (AUC), and testing time.
Main Results:
- The proposed CNN model achieved the highest accuracy (93.15%) in a seven-class classification task, outperforming VGG-16, VGG-19, and Inception-v3.
- The CNN model demonstrated strong performance across various multiclass classifications, with accuracies ranging from 96% to 98%.
- The CNN model exhibited shorter training and testing times compared to transfer learning models, indicating greater efficiency.
Conclusions:
- The developed CNN model provides an efficient and accurate method for diagnosing COVID-19 and other lung diseases from CXR images.
- This AI-driven approach has the potential to enhance real-time disease detection, aiding clinical decision-making and public health strategies.
- The study highlights the effectiveness of DL in medical image analysis for infectious disease identification.
