Guide-wire detecting using a modified cascade classifier in interventional radiology

Area of Science:

• Medical Imaging
• Surgical Technology
• Computer-Aided Surgery

Background:

• Endovascular surgery is increasingly used for cardiovascular diseases (CVDs).
• Guide-wires are essential tools in these procedures, requiring accurate tracking.
• Existing guide-wire tracking methods often involve complex modeling and high computational costs.

Purpose of the Study:

• To present a novel, automated guide-wire detection approach for endovascular surgery.
• To overcome limitations of existing methods, such as manual annotation and high complexity.
• To improve the accuracy and efficiency of guide-wire tracking during cardiovascular interventions.

Main Methods:

• A cascade classifier is employed for guide-wire detection, eliminating the need for explicit guide-wire modeling.
• The algorithm incorporates guide-wire motion direction to enhance detection accuracy.
• The approach is validated on a dataset of 349 frames.

Main Results:

• The proposed method achieves a mean tracking accuracy exceeding 95%.
• The algorithm demonstrates effective guide-wire detection under arbitrary motion.
• The approach significantly improves detection accuracy by considering motion direction.

Conclusions:

• The developed guide-wire detection algorithm is effective and accurate for endovascular procedures.
• This method offers an automated and computationally efficient alternative to existing tracking techniques.
• The findings support the potential of this approach to enhance surgical guidance in cardiovascular interventions.