Related Experiment Video
Updated: Aug 13, 2025
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
Lung and Infection CT-Scan-Based Segmentation with 3D UNet Architecture and Its Modification
Mohammad Hamid Asnawi1, Anindya Apriliyanti Pravitasari1, Gumgum Darmawan1
1Department of Statistics, Faculty of Mathematics and Natural Sciences, Universitas Padjadjaran, Bandung 45363, Indonesia.
Insights
The 3D UNet model achieved high accuracy in segmenting lungs and COVID-19 infections from CT scans. This AI-driven approach aids in faster diagnosis and assessing disease severity.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- COVID-19, declared a Public Health Emergency of International Concern, necessitates rapid and accurate diagnostic tools.
- Chest radiography, including CT scans, is crucial for COVID-19 diagnosis, but effective segmentation of affected areas is challenging.
- Existing segmentation algorithms require evaluation for their efficacy in segmenting lung and infection regions in COVID-19 CT scans.
Purpose of the Study:
- To evaluate the performance of the 3D UNet architecture and its modifications (3D ResUNet, 3D VGGUNet, 3D DenseUNet) for segmenting lung and infection areas in COVID-19 CT scans.
- To compare the effectiveness of binary-class (lung segmentation) and multi-class (lung and infection segmentation) approaches using these 3D segmentation models.
- To determine the optimal 3D segmentation model for improving COVID-19 diagnosis and severity assessment.
Main Methods:
- CT scan datasets were preprocessed using min-max scaling and Contrast Limited Adaptive Histogram Equalization (CLAHE).
- Four 3D deep learning segmentation architectures were implemented: 3D UNet, 3D ResUNet, 3D VGGUNet, and 3D DenseUNet.
- Model performance was evaluated on both binary-class (lung segmentation) and multi-class (lung and infection segmentation) tasks using IoU, Dice scores, and accuracy metrics.
Main Results:
- The original 3D UNet model outperformed the modified architectures in both binary and multi-class segmentation tasks.
- For binary-class segmentation, 3D UNet achieved IoU of 94.32%, Dice score of 97.05%, and accuracy of 99.37%.
- For multi-class segmentation, 3D UNet achieved IoU of 81.58%, Dice score of 88.61%, and accuracy of 98.78%, demonstrating its effectiveness in segmenting both lung and infection areas.
Conclusions:
- The 3D UNet architecture demonstrates superior performance for segmenting lung and COVID-19 infection regions in CT scans compared to its modifications.
- Accurate 3D segmentation of infection areas can significantly aid medical personnel in diagnosing COVID-19 and assessing disease severity.
- Further development and application of 3D UNet for medical image analysis hold promise for improving diagnostic workflows in infectious diseases.
Abstract:
COVID-19 is the disease that has spread over the world since December 2019. This disease has a negative impact on individuals, governments, and even the global economy, which has caused the WHO to declare COVID-19 as a PHEIC (Public Health Emergency of International Concern). Until now, there has been no medicine that can completely cure COVID-19. Therefore, to prevent the spread and reduce the negative impact of COVID-19, an accurate and fast test is needed. The use of chest radiography imaging technology, such as CXR and CT-scan, plays a significant role in the diagnosis of COVID-19. In this study, CT-scan segmentation will be carried out using the 3D version of the most recommended segmentation algorithm for bio-medical images, namely 3D UNet, and three other architectures from the 3D UNet modifications, namely 3D ResUNet, 3D VGGUNet, and 3D DenseUNet. These four architectures will be used in two cases of segmentation: binary-class segmentation, where each architecture will segment the lung area from a CT scan; and multi-class segmentation, where each architecture will segment the lung and infection area from a CT scan. Before entering the model, the dataset is preprocessed first by applying a minmax scaler to scale the pixel value to a range of zero to one, and the CLAHE method is also applied to eliminate intensity in homogeneity and noise from the data. Of the four models tested in this study, surprisingly, the original 3D UNet produced the most satisfactory results compared to the other three architectures, although it requires more iterations to obtain the maximum results. For the binary-class segmentation case, 3D UNet produced IoU scores, Dice scores, and accuracy of 94.32%, 97.05%, and 99.37%, respectively. For the case of multi-class segmentation, 3D UNet produced IoU scores, Dice scores, and accuracy of 81.58%, 88.61%, and 98.78%, respectively. The use of 3D segmentation architecture will be very helpful for medical personnel because, apart from helping the process of diagnosing someone with COVID-19, they can also find out the severity of the disease through 3D infection projections.
