Neural-network-based accelerated liver water T1 mapping

Motivation: Reliable wT1 mapping continuous inversion-recovery Look-Locker (CIR-LL) requires a certain recovery duration, which results in extended acquisition times per slice and limits the liver coverage within one breath-hold. Goal(s): To reduce the required recovery duration for reliable liver wT1 mapping. Approach: We propose to combine CIR-LL with a neural network, which is trained to conduct wT1 mapping based on the SVD of a reduced recovery duration. Results: The proposed time-efficient wT1 liver mapping method provides consistent wT1 values in phantom and in vivo measurements in both volunteers and patients, while preserving lesion detection. Impact: A neural network-based liver wT1 mapping technique that shortens acquisition times using CIR-LL without sacrificing the quality of quantitative reconstruction results, and holds promise for further mapping applications.