194 Evaluating a novel AI tool for automated measurement of the aortic root and valve from cardiac magnetic resonance imaging

Introduction This study aimed to evaluate the performance of 'AIATELLA,' an Artificial Intelligence (AI) system, in interpreting Cardiac Magnetic Resonance (CMR) images for aortic valve and root measurements during different cardiac cycle phases, comparing its accuracy and efficiency against senior National Health Service (NHS) cardiologists. Method The analysis was conducted by comparing the AI-generated measurements of the aortic valve and root across various cardiac cycle phases with those assessed by experienced NHS cardiologists. The study utilised a comprehensive dataset of CMR images, employing the Intraclass Correlation Coefficient (ICC) as the primary statistic for measuring agreement between the AI and cardiologist assessments. Results The comparison revealed a high level of agreement between 'AIATELLA' AI assessments and those made by senior NHS cardiologists, with an overall ICC of 0.98. Notably, the AI demonstrated the ability to match the accuracy of cardiologist measurements while delivering results more than 100 times faster, showcasing a significant improvement in efficiency without compromising accuracy. Conclusion The findings of this study underscore the significant potential of AI, particularly 'AIATELLA,' to enhance the interpretation of CMR images, thereby potentially revolutionising cardiac care. The high degree of accuracy, coupled with the dramatically increased speed of analysis, indicates that AI can be a valuable tool in cardiac diagnostics, offering a promising solution to the challenges of time-intensive, costly, and variable clinician-based assessments. This study highlights the crucial role that AI can play in advancing cardiac care, paving the way for more efficient, accurate, and accessible diagnostics in the future. Conflict of Interest None