A Hybrid Catheter Localisation Framework in Echocardiography Based on Electromagnetic Tracking and Deep Learning...

Area of Science:

• Interventional Cardiology
• Medical Imaging
• Artificial Intelligence

Background:

• Interventional cardiology utilizes minimally invasive techniques for cardiovascular diseases.
• Ultrasound imaging (echocardiography) is crucial for monitoring catheter placement.
• Accurate medical device localization is vital for procedural safety and reliability.

Purpose of the Study:

• To develop an automatic hybrid framework for precise catheter tip localization.
• To overcome limitations of standalone external device tracking and image-based tracking methods.

Main Methods:

• A hybrid framework integrating an electromagnetic tracking system (North Digital Inc) and a deep learning-based ultrasound image analysis (UNet) was developed.
• The electromagnetic tracking system provided precise external localization of the catheter tip.
• UNet performed automatic semantic segmentation of the catheter tip in ultrasound images.

Main Results:

• The hybrid framework successfully combined precise external localization with automatic image-based segmentation.
• This approach offers a novel solution for identifying moving medical devices in low-resolution ultrasound images.
• The integrated system enhances the accuracy and automation of catheter tip localization.

Conclusions:

• The proposed hybrid localization framework effectively integrates electromagnetic tracking and deep learning for interventional cardiology.
• This method improves the safety and reliability of catheter-based procedures by accurately identifying the catheter tip.
• The framework presents a significant advancement for real-time device tracking in challenging ultrasound environments.