Model-based image reconstruction for dynamic cardiac perfusion MRI from sparse data

The paper presents a novel approach for dynamic magnetic resonance imaging (MRI) cardiac perfusion image reconstruction from sparse k-space data. It formulates the reconstruction problem in an inverse-methods setting. Relevant prior information is incorporated via a parametric model for the perfusion process. This wealth of prior information empowers the proposed method to give high-quality reconstructions from very sparse k-space data. The paper presents reconstruction results using both Cartesian and radial sampling strategies using data simulated from a real acquisition. The proposed method produces high-quality reconstructions using 14% of the k-space data. The model-based approach can potentially greatly benefit cardiac myocardial perfusion studies as well as other dynamic contrast-enhanced MRI applications including tumor imaging.