Spectro-Dynamic MRI: Characterizing Mechanical Systems on a Millisecond Scale

Abstract

Measuring in vivo dynamics can yield valuable information for studying the functioning of the cardiovascular or the musculoskeletal system and for the diagnosis of related diseases. MRI is a powerful medical imaging modality, but it shows severe limitations when dealing with motion at high spatial and temporal resolutions. In this work, a method called spectro-dynamic MRI is proposed, which can identify dynamical information directly from k-space data. It combines a measurement model, relating the measured data in k-space to the displacement fields, and a dynamical model, introducing prior knowledge about the dynamics of a system. The data sampling process is tailored to compute spatial and temporal derivatives in the spectral domain at a high temporal resolution. Preliminary results from four simple pendulum setups for which the dynamics are explicitly known show that spectro-dynamic MRI can estimate motion fields from heavily undersampled data on a millisecond timescale. Furthermore, the length of the pendula and the stiffness of the spring can be identified as the dynamical system's parameters, giving additional information about the systems under investigation.