Context-Driven Active Contour (CDAC): A Novel Medical Image Segmentation Method Based on Active Contour and...

Area of Science:

• Medical Imaging
• Radiology
• Artificial Intelligence in Medicine

Background:

• Lung diseases such as COPD and pulmonary fibrosis present significant global health challenges.
• Computed tomography (CT) imaging is crucial for diagnosing and managing lung conditions.
• Traditional image segmentation methods struggle with the anatomical and pathological complexities of lung CT scans.

Purpose of the Study:

• To introduce a novel segmentation method, Context-Driven Active Contour (CDAC), for improved lung CT analysis.
• To address the limitations of existing segmentation techniques in handling anatomical variability and complex pathologies.
• To enhance the precision of computer-aided diagnostic (CAD) systems for lung disease management.

Main Methods:

• Developed the Context-Driven Active Contour (CDAC) method, integrating active contour models (ACMs) with contextual analysis.
• Utilized image embeddings and expert annotations to provide contextual information for segmentation refinement.
• Incorporated contextual attention force (CAF) and contextual balloon force (CBF) for robust contour adaptation.
• Evaluated CDAC on CT images of healthy lungs, COPD, and pulmonary fibrosis.

Main Results:

• CDAC achieved a Dice coefficient of 96.8% for healthy lung segmentation.
• The method demonstrated 94.5% accuracy in segmenting lungs affected by COPD.
• A Jaccard Index of 92.3% was obtained for pulmonary fibrosis segmentation.
• CDAC showed effectiveness and adaptability across different lung conditions.

Conclusions:

• CDAC offers a significant advancement in lung CT image segmentation.
• The integration of contextual information enhances segmentation precision and robustness.
• CDAC shows promise for improving the performance of computer-aided diagnostic (CAD) systems for lung diseases.