Explainable multi-instance and multi-task learning for COVID-19 diagnosis and lesion segmentation in CT images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Coronavirus Disease 2019 (COVID-19) remains a global pandemic.
• Chest CT scans are crucial for COVID-19 screening, offering detailed pathological information.
• Current automated methods often treat diagnosis and segmentation as separate tasks, lacking clinical explainability.

Purpose of the Study:

• To develop a unified, explainable multi-task learning framework for simultaneous COVID-19 diagnosis and multi-lesion segmentation in CT images.
• To improve the accuracy and clinical interpretability of automated COVID-19 detection and analysis.
• To enable severity assessment and lesion quantification for comprehensive patient status evaluation.

Main Methods:

• Development of an explainable multi-instance multi-task network.
• Simultaneous learning of task-related features with adaptive weights for diagnosis and segmentation.
• Integration of lesion highlighting for explicable diagnostic results.

Main Results:

• The proposed framework achieved superior performance in both COVID-19 diagnosis and multi-lesion segmentation compared to existing methods.
• The model demonstrated enhanced accuracy in identifying and segmenting lesions within lung CT images.
• The explainable nature of the network provided clinically relevant evidence for diagnoses.

Conclusions:

• The developed multi-task learning framework effectively integrates COVID-19 diagnosis and CT image segmentation.
• The explainable AI approach offers improved accuracy and interpretability for automated analysis of COVID-19.
• This method holds promise for enhanced clinical decision-making in managing COVID-19 patients.