Synthetic Contrast-Free LGE via Diffusion-Based Framework in Acute MI for Image Quality and Quantitative Scar...

Area of Science:

• Cardiovascular Imaging
• Artificial Intelligence in Medicine
• Medical Image Generation

Background:

• Late gadolinium enhancement (LGE) is crucial for assessing myocardial scar but requires contrast agents.
• Developing contrast-free methods for LGE-like imaging is essential for patient safety and accessibility.
• Accurate scar quantification in acute myocardial infarction (AMI) remains a clinical challenge.

Purpose of the Study:

• To develop and evaluate a diffusion model-based framework for generating LGE-like images without contrast agents.
• To assess the subjective and objective image quality of synthetic native enhancement (SNE) images.
• To determine the clinical utility of SNE images for quantifying myocardial scar in AMI patients.

Main Methods:

• A multisequence guided diffusion model was developed to generate SNE images from cine MRI and T2-STIR sequences.
• Data from 331 AMI patients across three centers were retrospectively analyzed.
• Image quality was assessed using Likert scoring and contrast ratio analysis; scar quantification used the full-width at half-maximum method.

Main Results:

• The multisequence guided diffusion model produced SNE images with superior visual quality and data distribution alignment compared to other generative models.
• SNE images showed significantly higher quality and improved contrast ratios versus standard LGE.
• SNE demonstrated strong agreement with LGE for scar size (R=0.839 internal, R=0.816 external) and transmurality (R=0.792 internal, R=0.758 external) with minimal bias.

Conclusions:

• Synthetic native enhancement (SNE) images generated by the diffusion model show strong agreement with LGE for quantitative scar assessment in AMI.
• The SNE technique offers comparable or improved image quality without the need for contrast agents.
• This contrast-free approach holds promise for accurate and accessible myocardial scar quantification.