X-ray coronary centerline extraction based on C-UNet and a multifactor reconnection algorithm.

Area of Science:

• Medical Imaging
• Cardiovascular Research
• Artificial Intelligence in Medicine

Background:

• Accurate coronary artery centerline extraction is vital for diagnosing stenosis, lesion detection, and surgical navigation.
• Challenges in X-ray coronary angiography include complex backgrounds, low signal-to-noise ratios, and intricate vascular structures.

Purpose of the Study:

• To develop an automated and accurate method for extracting coronary artery centerlines from X-ray coronary angiography images.
• To address the limitations of existing methods in complex clinical scenarios.

Main Methods:

• A novel centerline extraction method combining a U-Net based deep learning network (C-UNet) with a residual network.
• A multifactor centerline reconnection algorithm leveraging geometric characteristics of blood vessels.

Main Results:

• The proposed method demonstrates effectiveness through qualitative and quantitative evaluations.
• High precision, recall, and F1_Score indicate accurate coronary artery centerline extraction.

Conclusions:

• The developed method accurately extracts coronary artery centerlines from X-ray angiography.
• The approach improves both the accuracy and continuity of extracted centerlines, aiding clinical applications.