JCS: An Explainable COVID-19 Diagnosis System by Joint Classification and Segmentation.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Diseases

Background:

• The COVID-19 pandemic necessitates rapid and accurate diagnostic tools.
• Reverse Transcription Polymerase Chain Reaction (RT-PCR) tests have limitations in sensitivity.
• Chest CT scans offer high sensitivity for early COVID-19 detection but are time-consuming.

Purpose of the Study:

• To develop a real-time and explainable Joint Classification and Segmentation (JCS) system for COVID-19 diagnosis using chest CT scans.
• To create a large-scale dataset (COVID-CS) for training and evaluating AI models for COVID-19 chest CT analysis.

Main Methods:

• Development of a novel Joint Classification and Segmentation (JCS) system.
• Construction of the COVID-19 Classification and Segmentation (COVID-CS) dataset, including 144,167 chest CT images and detailed annotations.
• Training the JCS system on the COVID-CS dataset with pixel-level labels for opacifications.

Main Results:

• The JCS system achieved an average sensitivity of 95.0% and specificity of 93.0% for COVID-19 classification.
• The system obtained a Dice score of 78.5% for lung opacification segmentation.
• The developed system demonstrates efficiency and accuracy in real-time COVID-19 chest CT diagnosis.

Conclusions:

• The JCS system provides an efficient and accurate solution for COVID-19 diagnosis from chest CT scans.
• The COVID-CS dataset and JCS system can aid in the early detection and management of COVID-19.
• This AI-driven approach complements traditional diagnostic methods, offering faster and more sensitive analysis.