US Patent 11181598 Multi-contrast MRI image reconstruction using machine learning

A computer-implemented method for reconstructing a MRI image, including: receiving a plurality of MRI measurement data sets ƒ1 to ƒN, wherein each data set is acquired from an examination object based on a different MRI protocol of an MRI system; receiving MRI images u10 to uN0 corresponding to the MRI measurement data sets ƒ1 to ƒN; applying, in at least a first step GD1, trained functions to the MRI images u10 to uN0, using a neural network and a forward-sampling operator, wherein at least one output MRI image uT is generated; and providing the at least one output MRI image uT, wherein the forward-sampling operator determines an agreement between at least one MRI image u10 to uN0 and the corresponding MRI measurement data set ƒ1 to ƒN.