Diving into Extended Phase Graph-based Deep Learning for accurate T2 mapping with PENGUIN

Model-based deep learning approaches have shown promising results to accelerate T2 relaxometry, but most adopt a pure exponential curve to model the signal, which does not account for indirect and stimulated echoes. A PhasE graph sigNal and Gradients QUantitative Inference MachiNe (PENGUIN) is proposed, which implements a dictionary of pre-calculated echo-modulation curves following the Extended Phase Graph (EPG) formulation and respective gradients as the inputs of a Recurrent Inference Machine to perform accurate T2 mapping from the reconstructed images. PENGUIN is 25-fold faster than a pattern recognition approach with a T2 dictionary step of 2 ms.