Vessel segmentation and catheter detection in X-ray angiograms using superpixels

Area of Science:

• Medical Imaging
• Cardiovascular Disease Research
• Image Processing

Background:

• Coronary artery disease (CAD) is a major global health concern.
• X-ray angiography (XRA) is a key diagnostic tool for CAD.
• Challenges in XRA image analysis include poor contrast, non-uniform illumination, and artifacts like catheters, hindering accurate diagnosis.

Purpose of the Study:

• To develop an automated method for coronary artery segmentation in XRA images.
• To enable accurate detection and tracking of catheters within XRA sequences.
• To extract coronary artery centerlines for improved diagnostic insights.

Main Methods:

• Utilized multi-scale superpixel analysis with a vesselness probability measure.
• Implemented a voting mechanism for initial segmentation, refined by orthogonal line detection on vessel ridges.
• Employed polynomial fitting for catheter detection and tracking across angiography frames.
• Applied image ridge detection for centerline extraction.

Main Results:

• The proposed method demonstrated superior performance in segmenting coronary arteries compared to a previous technique.
• Cardiologist assessments indicated 83% of images processed by the new method were rated good or excellent, versus 48% for the comparison.
• The new method achieved a 67% improvement in processing speed over the compared approach.

Conclusions:

• The developed method offers a robust solution for coronary artery segmentation and catheter management in XRA.
• This advancement has the potential to improve the accuracy and efficiency of CAD diagnosis.
• The technique provides reliable vessel segmentation and centerline extraction, aiding clinical decision-making.