Myocardial Infarction Detection with Incomplete Multi-View Data via Dual-Branch Gating Completion and Dirichlet...

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Imaging

Background:

• Electrocardiogram (ECG) for myocardial infarction (MI) detection is limited by noise and early diagnostic value.
• Echocardiography offers richer data but often has missing views due to clinical constraints.
• Current fusion methods struggle with adaptability and real-time application due to static or complex dynamic weighting.

Purpose of the Study:

• To develop an advanced view completion method for multi-view echocardiography in MI detection.
• To enhance the fusion of incomplete multi-view data for improved diagnostic performance.
• To overcome limitations of existing fusion techniques by introducing adaptive weighting.

Main Methods:

• A dual-branch gating structure combining Transformer and Graph Neural Network (GNN) for view completion.
• Transformer encoder for global temporal dependencies and GNN for local structural relationships.
• Uncertainty-driven dynamic weighted fusion strategy using Dirichlet distribution for adaptive weighting.

Main Results:

• The proposed method achieved 92.31% accuracy, 90.00% precision, and 100.00% specificity on the HMC-QU dataset.
• Demonstrated superior performance compared to state-of-the-art models in multi-view MI detection.
• The adaptive fusion strategy effectively handled varying view quality and improved diagnostic confidence.

Conclusions:

• The novel AI approach significantly improves MI detection accuracy with incomplete echocardiography data.
• The dual-branch gating structure and dynamic fusion strategy offer a robust solution for real-world clinical scenarios.
• The method shows strong potential for clinical deployment, enhancing early and accurate MI diagnosis.