Joint MR T1 and T2* Parameter Mapping with Scan Specific Unsupervised Networks

We introduce a new method for joint T1 and T2* MR parameter mapping in MAPLE framework which takes advantage of multi-echo multi-flip angle (MEMFA) gradient echo data. It combines joint reconstruction of MEMFA data with a joint relaxation signal model to improve parameter estimation. The proposed method estimates T1 and T2* parameters jointly with enough flexibility to incorporate different standard and state-of-the-art reconstruction methods, including zero-shot self-supervised learning (ZS-SSL) reconstruction. It improves the reconstruction performance of the methods with learnable parameters by offering a physics-based additional regularization into their optimization process and exploiting spatio-temporal correlations.