A multiple catheter tips tracking method in X-ray fluoroscopy images by a new lightweight segmentation network and...

Area of Science:

• Medical Imaging
• Interventional Cardiology
• Computer Vision

Background:

• Percutaneous coronary intervention (PCI) relies on guiding catheters for accurate placement.
• X-ray fluoroscopy is crucial for visualizing catheter tips but suffers from noise and low contrast due to low radiation doses.
• Accurate tracking of the catheter tip is essential for dynamic coronary roadmap generation and overlap.

Purpose of the Study:

• To develop and validate a robust catheter tip tracking framework for X-ray fluoroscopy sequences.
• To improve the accuracy of dynamic coronary roadmap generation by precisely tracking catheter tip motion.
• To address the challenges of low signal-to-noise ratio and low contrast in X-ray fluoroscopy images.

Main Methods:

• A lightweight, efficient catheter tip segmentation network was developed, enhanced with a self-distillation training mechanism.
• Bayesian filtering was employed for post-processing to refine segmentation by incorporating temporal sequence information.
• The framework was designed to track multiple catheter tips by grouping segmentation results based on connectivity.

Main Results:

• The proposed framework demonstrated effective catheter tip segmentation and tracking in noisy, low-contrast X-ray fluoroscopy images.
• The self-distillation mechanism improved the efficiency and accuracy of the segmentation network.
• Bayesian filtering successfully refined tracking by leveraging temporal data, and multi-tip tracking capability was achieved.

Conclusions:

• The developed framework provides a reliable method for catheter tip tracking in X-ray fluoroscopy during PCI.
• This technology has the potential to enhance the precision and safety of interventional cardiology procedures.
• The framework's ability to handle challenging imaging conditions and track multiple tips offers significant clinical value.