Aapo Nummenmaa1, Matti S. Hamalainen1, Fa-Hsuan Lin1,2
1Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, USA; 2Institute for Biomedical Engineering, National Taiwan University, Taipei, 106, Taiwan
We propose a simple method for automatic regularization of dynamic magnetic resonance Inverse Imaging (InI). Regularization is interpreted in a Bayesian way, as a variance parameter of a Gaussian prior, and marginal likelihood is used to estimate these parameters. The proposed method is compared to the presently used ad hoc regularization of InI by using empirical data from a visual stimulation experiment. Possible extension of the method for dynamic modeling of the regularization parameters is discussed.