Related Experiment Video
Updated: Nov 21, 2025
08:05
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
14.5K
COVIDScreen: explainable deep learning framework for differential diagnosis of COVID-19 using chest X-rays
Rajeev Kumar Singh1, Rohan Pandey1, Rishie Nandhan Babu1
1Shiv Nadar University, NCR, Gautam Budh Nagar, India.
Neural Computing & Applications
|January 13, 2021
Summary
A novel deep learning model using chest X-rays offers rapid COVID-19 diagnosis. This AI system achieves high accuracy, aiding in quick patient triage when RT-PCR tests are unavailable.
Area of Science:
- Artificial Intelligence
- Medical Imaging
- Deep Learning
Background:
- The COVID-19 pandemic presents significant global challenges, necessitating rapid diagnostic tools.
- Limitations in RT-PCR testing, including shortages and delays, highlight the need for alternative methods.
- Accurate and swift patient triaging is crucial for controlling the spread of COVID-19 infections.
Purpose of the Study:
- To develop a deep learning-based system for rapid COVID-19 patient triaging using chest X-rays.
- To enhance the reliability and speed of COVID-19 diagnosis in resource-constrained settings.
- To improve model performance and generalizability through advanced techniques like pruning and explainability.
Main Methods:
- Image enhancement and segmentation techniques applied to chest X-ray images.
- A modified stacked ensemble model integrating four CNN base-learners and a Naive Bayes meta-learner.
- Implementation of an effective pruning strategy to optimize model complexity and performance.
- Utilization of Grad-CAM for explainability and Generative Adversarial Networks (GANs) for synthetic data generation.
Main Results:
- The proposed deep learning model achieved high diagnostic accuracy (98.67%) and a Kappa score of 0.98.
- Excellent F-1 scores were obtained: 100% for COVID-19, 98% for normal, and 98% for pneumonia classifications.
- The system demonstrated superior performance compared to existing methods on standard datasets.
- GANs effectively generated realistic synthetic COVID-19 chest X-ray samples to augment limited training data.
Conclusions:
- The developed deep learning solution offers a rapid and accurate method for COVID-19 detection using chest X-rays.
- This AI system can serve as a valuable tool for preliminary patient evaluation, complementing traditional diagnostic methods.
- The integration of explainability and synthetic data generation enhances the trustworthiness and robustness of the AI model.
