Related Experiment Video
Updated: Jun 28, 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
14.2K
DDA-SSNets: Dual decoder attention-based semantic segmentation networks for COVID-19 infection segmentation and
Anandbabu Gopatoti1, Ramya Jayakumar1, Poornaiah Billa2
1Department of Electronics and Communication Engineering, Hindusthan College of Engineering and Technology, Coimbatore, Tamil Nadu, India.
Journal of X-Ray Science and Technology
|April 12, 2024
Summary
Deep learning models accurately detect and grade COVID-19 lung infections using chest X-rays. Dual Decoder Attention-SegNet and GADCNet show superior performance in segmentation and classification tasks.
Area of Science:
- Artificial Intelligence in Medical Imaging
- Deep Learning for Disease Diagnosis
- Radiology and Medical Image Analysis
Background:
- Accurate diagnosis and staging of COVID-19 are crucial for effective treatment.
- Existing diagnostic and staging methods for COVID-19 infection require improvement.
- Advanced deep learning approaches are needed to assist radiologists in identifying and quantifying lung infections associated with COVID-19.
Purpose of the Study:
- To develop and evaluate deep learning-based models for classifying and quantifying COVID-19 related lung infections.
- To improve the accuracy of COVID-19 detection and severity grading using chest X-ray (CXR) images.
- To introduce novel deep learning architectures for enhanced medical image analysis in the context of COVID-19.
Main Methods:
- Proposed Dual Decoder Attention-based Semantic Segmentation Networks (DDA-SSNets), including DDA-UNet and DDA-SegNet, for segmenting lung lobes and infections in CXRs.
- Utilized segmentation outputs to grade the severity of COVID-19 infection in lung lobes.
- Developed a Genetic Algorithm-based Deep Convolutional Neural Network classifier (GADCNet) for classifying COVID-19 versus non-COVID-19 in extracted regions of interest.
Main Results:
- DDA-SegNet demonstrated superior segmentation performance with average BCSSDC scores of 99.53% (lung lobes) and 99.97% (infection).
- The combined DDA-SegNet and GADCNet classifier achieved excellent classification accuracy with an average BCCAC of 99.98%.
- Comparative analysis showed DDA-SegNet outperformed DDA-UNet in segmentation and GADCNet with DDA-UNet achieved 99.92% BCCAC.
Conclusions:
- The proposed DDA-SegNet significantly enhances the segmentation of lung lobes and COVID-19 infected regions in CXRs.
- The DDA-SegNet model provides improved severity grading of COVID-19 lung infections compared to DDA-UNet.
- The GADCNet classifier, particularly when combined with DDA-SegNet, demonstrates high accuracy in classifying CXRs for COVID-19 detection.
