Related Experiment Video
Updated: Aug 23, 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
14.3K
A novel deep learning-based method for COVID-19 pneumonia detection from CT images
Ju Luo1, Yuhao Sun2, Jingshu Chi1
1Third Xiangya Hospital, Central South University, NO.138, Tongzipo Road, Changsha, 410013, Hunan, China.
BMC Medical Informatics and Decision Making
|November 3, 2022
Summary
A deep learning model shows high accuracy in detecting COVID-19 pneumonia from chest CT scans, assisting radiologists. This AI tool significantly reduces diagnosis time compared to human experts.
Area of Science:
- Artificial Intelligence in Medical Imaging
- Radiology and Diagnostic Imaging
- Infectious Disease Diagnostics
Background:
- Reverse transcription polymerase chain reaction (RT-PCR) has limited sensitivity (60-70%) for COVID-19 diagnosis.
- Chest CT is crucial for diagnosing COVID-19 pneumonia but relies heavily on radiologist expertise.
- Limitations in RT-PCR sensitivity and radiologist dependency highlight the need for advanced diagnostic tools.
Purpose of the Study:
- To develop a deep learning (AI) model to aid radiologists in identifying COVID-19 pneumonia.
- To enhance the accuracy and efficiency of COVID-19 pneumonia detection using artificial intelligence.
- To create a decision-support tool for diagnosing COVID-19 pneumonia.
Main Methods:
- A deep learning model was designed using U-Net and ResNet-50 architectures.
- The model was trained and validated on a large dataset of chest CT images (normal, Community-Acquired Pneumonia (CAP), and COVID-19).
- The AI model's diagnostic performance was compared against radiologists of varying experience levels.
Main Results:
- The AI model achieved high diagnostic sensitivity for normal cases (98.03%), CAP (89.28%), and COVID-19 (92.15%) in the test set.
- Overall diagnostic accuracy reached 93.84% in the test set and 92.86% in the validation set.
- The AI model significantly outperformed all radiologists in terms of diagnostic time and showed comparable accuracy to expert radiologists.
Conclusions:
- The developed AI model demonstrates robust decision-making capabilities for detecting COVID-19 pneumonia.
- This AI tool has the potential to significantly assist clinicians in diagnosing COVID-19 pneumonia.
- The AI model offers a promising solution to improve diagnostic efficiency and accuracy in the context of COVID-19.
