A Collaborative Multimodal Learning-Based Framework for COVID-19 Diagnosis.

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Public Health

Background:

• The COVID-19 pandemic created a global health crisis, straining medical resources and highlighting the need for efficient diagnostic tools.
• Automated COVID-19 diagnosis systems are crucial for assisting clinicians and reducing the burden of image interpretation, especially with evolving viral mutations.
• Data scarcity and privacy concerns at individual institutions hinder the development of robust diagnostic models by limiting access to diverse datasets.

Purpose of the Study:

• To propose a novel, privacy-preserving, cross-site framework for diagnosing COVID-19 using multimodal data.
• To effectively utilize heterogeneous data from multiple sources while ensuring patient privacy.
• To improve the accuracy and efficiency of COVID-19 diagnosis in clinical settings.

Main Methods:

• Development of a privacy-preserving cross-site framework integrating multimodal data from various institutions.
• Introduction of a Siamese branched network as the backbone architecture to analyze relationships within heterogeneous samples.
• Implementation of semisupervised learning capabilities within the network to handle varied data modalities and improve task-specific training.

Main Results:

• The proposed framework demonstrated significant performance improvements compared to existing state-of-the-art methods.
• Extensive simulations on real-world datasets validated the effectiveness of the privacy-preserving approach.
• The model successfully handled semisupervised inputs across multiple modalities, enhancing diagnostic accuracy.

Conclusions:

• The developed framework offers a viable solution for privacy-preserving, collaborative COVID-19 diagnosis using multimodal data.
• This approach effectively addresses data limitations and privacy restrictions in federated medical learning.
• The enhanced diagnostic capabilities hold significant promise for improving patient care and managing public health crises.