Related Experiment Video
Updated: Oct 2, 2025
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
Cascaded 3D UNet architecture for segmenting the COVID-19 infection from lung CT volume
Aswathy A L1, Vinod Chandra S S2
1Department of Computer Science, University of Kerala, Thiruvananthapuram, India. aswathysatheesh@keralauniversity.ac.in.
Insights
This study introduces a novel two-stage 3D UNet deep learning model for segmenting lung infections in COVID-19 patients using CT scans. The model accurately identifies infected areas, aiding diagnosis and prognosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Pulmonology
Background:
- The COVID-19 pandemic necessitates accurate diagnostic tools for timely intervention.
- Computed Tomography (CT) scans are crucial for visualizing lung involvement in COVID-19 patients.
- Deep learning offers potential for automated analysis of medical images.
Purpose of the Study:
- To develop and evaluate a deep learning model for segmenting lung parenchyma and infected areas in COVID-19 CT scans.
- To automate the process of identifying COVID-19 related lung abnormalities.
- To provide clinicians with a tool for efficient analysis of CT volumes.
Main Methods:
- A two-stage cascaded 3D UNet architecture was employed.
- The first 3D UNet segmented lung parenchyma from preprocessed CT volumes.
- The second 3D UNet identified infected regions within the segmented lung parenchyma.
Main Results:
- The lung parenchyma segmentation achieved 93.47% sensitivity, 98.64% specificity, 98.07% accuracy, and a 92.46% dice score.
- Lung infection segmentation yielded 83.33% sensitivity, 99.84% specificity, 99.20% accuracy, and an 82% dice score.
- The model automates analysis without requiring manual lung parenchyma labeling for each patient.
Conclusions:
- The proposed two-stage 3D UNet effectively segments lung parenchyma and COVID-19 infected areas from CT scans.
- This deep learning approach enhances diagnostic capabilities for COVID-19 lung infections.
- The automated segmentation can assist clinicians in patient management and prognosis assessment.
Abstract:
World Health Organization (WHO) declared COVID-19 (COronaVIrus Disease 2019) as pandemic on March 11, 2020. Ever since then, the virus is undergoing different mutations, with a high rate of dissemination. The diagnosis and prognosis of COVID-19 are critical in bringing the situation under control. COVID-19 virus replicates in the lungs after entering the upper respiratory system, causing pneumonia and mortality. Deep learning has a significant role in detecting infections from the Computed Tomography (CT). With the help of basic image processing techniques and deep learning, we have developed a two stage cascaded 3D UNet to segment the contaminated area from the lungs. The first 3D UNet extracts the lung parenchyma from the CT volume input after preprocessing and augmentation. Since the CT volume is small, we apply appropriate post-processing to the lung parenchyma and input these volumes into the second 3D UNet. The second 3D UNet extracts the infected 3D volumes. With this method, clinicians can input the complete CT volume of the patient and analyze the contaminated area without having to label the lung parenchyma for each new patient. For lung parenchyma segmentation, the proposed method obtained a sensitivity of 93.47%, specificity of 98.64%, an accuracy of 98.07%, and a dice score of 92.46%. We have achieved a sensitivity of 83.33%, a specificity of 99.84%, an accuracy of 99.20%, and a dice score of 82% for lung infection segmentation.
