Related Experiment Video
Updated: Dec 13, 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
14.6K
Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-rays.
Sivaramakrishnan Rajaraman1, Jen Siegelman2, Philip O Alderson3
1Lister Hill National Center for Biomedical Communications, National Library of Medicine, Bethesda, MD 20894 USA.
IEEE Access : Practical Innovations, Open Solutions
|August 4, 2020
Summary
Deep learning models trained on chest X-rays accurately detect COVID-19 pulmonary manifestations. Iterative pruning and ensemble strategies achieved 99.01% accuracy in identifying viral abnormalities.
Area of Science:
- Artificial Intelligence
- Medical Imaging
- Deep Learning
Background:
- COVID-19, caused by SARS-CoV-2, presents pulmonary manifestations detectable via chest X-rays (CXRs).
- Accurate and efficient detection of COVID-19 on CXRs is crucial for timely diagnosis and patient management.
Purpose of the Study:
- To develop and evaluate an AI model for detecting COVID-19 pulmonary abnormalities using CXRs.
- To enhance model performance and efficiency through iterative pruning and ensemble learning.
Main Methods:
- Utilized custom and ImageNet-pretrained convolutional neural networks (CNNs) trained on CXR datasets.
- Employed knowledge transfer and fine-tuning for classifying normal, bacterial pneumonia, and COVID-19 CXRs.
- Applied iterative pruning to reduce model complexity and ensemble strategies to combine predictions.
Main Results:
- The best-performing pruned models, combined via weighted averaging, achieved 99.01% accuracy.
- An area under the curve (AUC) of 0.9972 was obtained in detecting COVID-19 findings.
- The integrated approach of knowledge transfer, pruning, and ensembling improved prediction accuracy.
Conclusions:
- Iteratively pruned deep learning model ensembles demonstrate high efficacy in detecting COVID-19 on CXRs.
- This AI-driven approach offers a promising tool for rapid COVID-19 screening using chest radiographs.
- The model's efficiency and accuracy suggest potential for clinical adoption in disease detection.
