Coronary Calcium Detection using 3D Attention Identical Dual Deep Network Based on Weakly Supervised Learning.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Cardiology

Background:

• Coronary artery calcium (CAC) indicates advanced subclinical coronary artery disease, predicting future cardiovascular events.
• Manual CAC detection from CT scans is the gold standard but is labor-intensive and impractical for large populations.

Purpose of the Study:

• To develop an automated deep learning model for efficient and accurate CAC detection.
• To utilize weakly supervised attention mechanisms on longitudinal non-contrast CT scans for CAC identification.

Main Methods:

• Proposed the attention identical dual network (AID-Net) incorporating 3D attention mechanisms for enhanced classification.
• Employed weakly supervised learning using only per-scan labels, reducing annotation burden.
• Integrated 3D Gradient-weighted Class Activation Mapping (Grad-CAM) for model interpretability.

Main Results:

• The AID-Net achieved a classification accuracy of 0.9272 and an AUC of 0.9627 on a dataset of 5075 non-contrast chest CT scans.
• Demonstrated superior performance compared to baseline methods.
• The model effectively leverages longitudinal scan data for improved detection.

Conclusions:

• The AID-Net provides a highly accurate and efficient automated solution for CAC detection.
• This deep learning approach facilitates large-scale screening for subclinical coronary artery disease.
• The interpretability of the model aids in clinical trust and understanding.