A region of interest prediction method for real-time online dynamic magnetic resonance imaging

Previously, online dynamic magnetic resonance imaging (dMRI) method based on motion estimate and motion compensation (ME/MC) technique was proposed, which reconstruct each frame immediately after its data have been acquired. ME/MC technique was used to predict current frame and can provide more sparser difference signal between frames for compressed sensing (CS) based reconstruction. In this study, a novel ME/MC prediction method based on region of interest (ROI) was proposed. ME/MC prediction is only applied on dramatically changing regions (ROI region), with the static regions predicted directly from previous frame. Retrospective study with cardiac cine data shown that proposed online dMRI scheme based on MPOD reconstruction with ME/MC prediction in ROI can not only accelerate the online reconstruction, but also provide higher image quality than that of the method based ME/MC technique.