A Tensor-Based Catheter and Wire Detection and Tracking Framework and Its Clinical Applications

Area of Science:

• Medical Imaging
• Image Processing
• Computer-Aided Diagnosis

Background:

• Catheters and wires are crucial in cardiac catheterization.
• Accurate detection of these instruments in fluoroscopic X-ray images is vital for clinical applications like motion compensation and 2D/3D co-registration.
• Detecting the full length and electrode positions of catheters/wires presents a significant challenge.

Purpose of the Study:

• To develop an automatic detection framework for catheters and wires in fluoroscopic X-ray images.
• To enable real-time tracking of these instruments for enhanced image-guidance.
• To validate the framework's performance in clinical applications.

Main Methods:

• Developed a framework based on path reconstruction from image tensors derived from a multiscale vessel enhancement filter.
• Utilized eigen direction vectors to detect catheters and wires as smooth paths.
• Implemented a real-time tracking method using a template generated from the detection process.

Main Results:

• Tested on 7,754 X-ray images with detection errors of 0.56 ± 0.28 mm for catheters and 0.68 ± 0.33 mm for guidewires.
• Achieved 2D target registration errors (TRE) of 1.8 mm ± 0.9 mm for motion compensation.
• Attained a TRE of 2.3 ± 0.9 mm for co-registration between 2D X-ray and 3D MRI models.

Conclusions:

• A novel, fully automatic detection framework for catheters and wires has been developed.
• The framework demonstrates high accuracy and effectiveness in clinical applications.
• This technology can significantly improve image-guidance systems in cardiac catheterization.