Skip to main content
2025 SCMR Conference Main banner
Final Program


Schedule at a Glance
Back

Rapid Fire Abstracts

Rapid Fire 06- Machine Learning and Artificial Intelligence III

Development and Validation of Automated Methods for Quantifying Late Gadolinium Enhancement in Cardiac MRI for Hypertrophic Cardiomyopathy (RF_FR_300)

Friday, January 31, 2025
12:00 PM – 12:10 PM East Coast USA Time
Location: Blue Room PreFunction
Background: Cardiac MRI plays a crucial role in assessing heart damage and fibrosis, particularly with Late Gadolinium Enhancement (LGE). However, manual analysis of LGE is both accurate and labor-intensive, requiring expertise. In response, we developed and compared two automated LGE quantification techniques, Auto Signal Threshold versus Reference Mean (ASTRM) and Auto Full Width at Half Maximum (AFWHM). Our goal was to assess their accuracy and efficiency compared to manual analysis.

Methods: STRM uses mean and standard deviation to identify a normal myocardial reference area, with a 6SD threshold in this study. FWHM selects an enhanced region based on its FWHM value. Myomics, an AI cardiac analysis software, automatically segments myocardial contours from LGE images. Then, normal regions for STRM and LGE areas for FWHM are automatically selected using proposed method, the LGE-enhanced areas from these selections are compared to manually selected reference regions by a 10-year experienced expert.

Results: The proposed method was evaluated in 50 patients with hypertrophic cardiomyopathy (HCM). Bland-Altman analysis revealed that both STRM and FWHM methods demonstrated over 96% of data points within the 95% confidence interval compared to expert manual selection reference areas, indicating a high level of agreement. Furthermore, automated analysis enabled simultaneous processing of multiple cases, resulting in a time savings of over 95% compared to manual assessment, highlighting its efficiency.

Conclusion: The results showed that both STRM and FWHM achieved high accuracy, with over 96% of data points agreeing with the manual selections. Additionally, automation significantly reduced analysis time compared to manual methods. Automated methods for analyzing cardiac imaging in LGE streamline the assessment of myocardial injury, providing fast, precise, and unbiased results. They improve workflow efficiency, maintain consistency in patient care, and optimize resource allocation. These automated tools could transform cardiovascular disease management, resulting in better patient outcomes and more cost-effective healthcare delivery.