Abstract
Deep learning-based methods have achieved prestigious performance formagnetic resonance imaging (MRI) reconstruction, enabling fast imaging for manyclinical applications. Previous methods employ convolutional networks to learnthe image prior as the regularization term. In quantitative MRI, the physicalmodel of nuclear magnetic resonance relaxometry is known, providing additionalprior knowledge for image reconstruction. However, traditional reconstructionnetworks are limited to learning the spatial domain prior knowledge, ignoringthe relaxometry prior. Therefore, we propose a relaxometry-guided quantitativeMRI reconstruction framework to learn the spatial prior from data and therelaxometry prior from MRI physics. Additionally, we also evaluated theperformance of two popular reconstruction backbones, namely, recurrentvariational networks (RVN) and variational networks (VN) with U- Net.Experiments demonstrate that the proposed method achieves highly promisingresults in quantitative MRI reconstruction.