Accelerating DT-CMR with Variable CAIPIRINHA Shifts and Artefact-Aware AI-based SMS reconstruction

Motivation: Simultaneous multi-slice (SMS) methods could reduce Diffusion-Tensor Cardiac MR (DT-CMR) scan times from over 60 minutes to clinically-feasible durations, but interslice leakage artefacts currently prevent practical use. Goal(s): Develop a method that enables SMS acceleration factors up to 3 while maintaining diffusion tensor map quality comparable to single-slice acquisition. Approach: We combined a novel variable CAIPIRINHA shift strategy with a dual-stream AI reconstruction method, evaluated through retrospective SMS simulations based on ex-vivo DT-CMR acquisitions. Results: Our method reduced DT-CMR map errors by up to 50% maintaining a 2.6× speed-up, with diffusion metrics closely matching non-accelerated acquisitions whereas standard SMS reconstruction showed significant deviations. Impact: By substantially reducing acquisition times while preserving diffusion map quality, this work helps bridge the gap towards clinical adoption of DT-CMR, advancing non-invasive assessment of cardiac microstructure for diagnosing and monitoring of cardiac disease.