Myocardial perfusion assessment using cardiac MRI allows non-invasive assessment of myocardial ischemia. In myocardial perfusion sequence, imaging is collected after a saturation pulse. An alternative approach based on steady-state imaging with radial sampling has been recently proposed. However, image reconstruction using compressed sensing in steady-state myocardial perfusion remains long and clinically not feasible. In this study, we sought to develop a deep learning-based image reconstruction platform for myocardial perfusion imaging.