Related Experiment Video
Updated: Oct 2, 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
14.4K
Optimized chest X-ray image semantic segmentation networks for COVID-19 early detection
Anandbabu Gopatoti1,2, P Vijayalakshmi1
1Department of Electronics and Communication Engineering, Hindusthan College of Engineering and Technology, Coimbatore, Tamil Nadu, India.
Journal of X-Ray Science and Technology
|February 25, 2022
Summary
This study enhances COVID-19 detection from chest X-rays using optimized deep learning semantic segmentation. Optimized SegNet achieved 98.08% accuracy, improving early disease identification from medical images.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Chest X-ray (CXR) radiography offers faster COVID-19 detection than PCR testing.
- Current deep learning models show limitations in accurately detecting COVID-19 from CXR images.
Purpose of the Study:
- To classify COVID-19 and normal patients using CXR images.
- To detect and label COVID-19 infected lung lobes via semantic segmentation networks.
Main Methods:
- Investigated SegNet, U-Net, and hybrid CNNs for semantic segmentation of infected lung lobes.
- Developed optimized networks (GWO SegNet, GWO U-Net, GWO hybrid CNN) using the grey wolf optimization (GWO) algorithm.
- Trained and validated models on a dataset of 2,572 COVID-19 CXR images.
Main Results:
- All optimized semantic segmentation networks exceeded 92% detection accuracy.
- Optimized SegNet demonstrated superior performance, achieving 98.08% accuracy in segmenting and classifying COVID-19 infected lung lobes.
- Outperformed optimized U-Net and hybrid CNN models.
Conclusions:
- Optimized deep learning networks show potential for objective and accurate COVID-19 identification.
- Semantic segmentation of CXR images can significantly aid in early COVID-19 detection.
- Further development of these AI models can improve diagnostic capabilities for respiratory diseases.
