Rapid 3D T1 mapping using deep learning-assisted Look-Locker inversion recovery MRI

Purpose: Conventional 3D Look-Locker inversion recovery (LLIR) T-1 mapping requires multi-repetition data acquisition to reconstruct images at different inversion times for T-1 fitting. To ensure B-1 robustness, sufficient time of delay (TD) is needed between repetitions, which prolongs scan time. This work proposes a novel deep learning-assisted LLIR MRI approach for rapid 3D T-1 mapping without TD. Theory and Methods: The proposed approach is based on the fact that T-1*, the effective T-1 in LLIR imaging, is independent of TD and can be estimated from both LLIR imaging with and without TD, while accurate conversion of T1* to T-1 requires TD. Therefore, deep learning can be used to learn the conversion of T1* to T-1, which eliminates the need for TD. This idea was implemented for inversion-recovery-prepared Golden-angel RAdial Sparse Parallel T-1 mapping (GraspT(1)). 39 GraspT(1) datasets with a TD of 6 s (GraspT(1)-TD6) were used for training, which also incorporates additional anatomical images. The trained network was applied for T-1 estimation in 14 GraspT(1) datasets without TD (GraspT(1)-TD0). The robustness of the trained network was also tested. Results: Deep learning-based T-1 estimation from GraspT(1)-TD0 is accurate compared to the reference. Incorporation of additional anatomical images improves the accuracy of T-1 estimation. The technique is also robust against slight variation in spatial resolution, imaging orientation and scanner platform. Conclusion: Our approach eliminates the need for TD in 3D LLIR imaging without affecting the T-1 estimation accuracy. It represents a novel use of deep learning towards more efficient and robust 3D LLIR T-1 mapping.

Automated summary

This work proposes a novel deep learning-assisted LLIR MRI approach for rapid 3D T-1 mapping without time delay, achieving accurate T-1 estimation.