A framework of myocardial bridge detection with x-ray angiography sequence

Area of Science:

• Cardiology
• Medical Imaging
• Artificial Intelligence

Background:

• Myocardial bridges are congenital abnormalities where heart muscle covers coronary arteries, potentially causing ischemia and sudden death.
• Automatic diagnosis of myocardial bridges is challenging.

Purpose of the Study:

• To develop a novel framework for automatic detection of myocardial bridges and vessel stenosis using x-ray angiography sequences.
• To improve diagnostic accuracy for coronary artery abnormalities.

Main Methods:

• A novel neural network combining CNNs and transformers for coronary vessel segmentation.
• Vessel tree description and inter-sequence information fusion.
• Automatic myocardial bridge detection based on vessel stenosis identification.

Main Results:

• Achieved Dice score of 0.917 and ASD of 1.39 for vessel segmentation.
• Attained 92.7% accuracy in detecting 262 stenoses.
• Reached 75% accuracy in myocardial bridge detection with multi-frame processing.

Conclusions:

• The proposed algorithm successfully automates the detection of stenosis and myocardial bridges.
• This framework offers a new approach for the automated diagnosis of coronary artery diseases.