A New Aortic Valve Calcium Scoring Framework for Automatic Calcification Detection in Echocardiography

Area of Science:

• Cardiology
• Medical Imaging
• Artificial Intelligence

Background:

• Aortic valve calcium scoring is crucial for managing aortic stenosis.
• Computed tomography (CT) is the current standard but involves radiation and high costs.
• Echocardiography is safer and cheaper but lacks objective, consistent measurements.

Purpose of the Study:

• To develop an automated, objective method for aortic valve calcium scoring using echocardiography.
• To improve cost-effectiveness and patient safety in aortic stenosis assessment.
• To overcome the subjectivity and variability of manual echocardiographic interpretations.

Main Methods:

• Utilized YOLOv5 for detecting the aorta region in echocardiography images.
• Developed a novel segmentation approach combining UNet and diffusion models for calcified areas.
• Validated the method on a dataset of 160 echocardiography images from 86 patients.

Main Results:

• Achieved 85.08% precision, 80.01% recall, and 71.13% Dice score for calcified area segmentation.
• Demonstrated a novel AI-driven approach for automated aortic valve calcium scoring.
• Showcased the potential for reducing costs and eliminating radiation exposure.

Conclusions:

• The proposed automated echocardiography-based method provides an objective and efficient tool for aortic valve calcium scoring.
• This AI-driven system can enhance diagnostic accuracy and patient safety in aortic stenosis management.
• Offers a viable, radiation-free alternative for frequent monitoring and risk assessment.