AI in CMR

This project aims to improve the efficiency of cardiac MRI (CMR) reporting workflows by integrating the output of multiple AI-driven processes into a unified pipeline. The system combines AI-based image post-processing (Sonic DL), automated segmentation of all CMR series (SHAIPS), draft report generation from quantitative measurements (HeartTalk), and patient-friendly report summarisation (MIRACLE).

The goal of this integrated pipeline is to halve CMR acquisition, analysis and reporting time from approximately 120 minutes to 60 minutes per scan while improving consistency and interpretability. The pipeline is currently under active development.