Baseline constrained reconstruction of DSC-MRI tracer kinetics from sparse fourier data

In order to assess brain perfusion, one of the available methods is the estimation of parameters such as cerebral blood flow (CBF), cerebral blood volume (CBV) and mean transit time (MTT) from Dynamic Susceptibility Contrast MRI (DSC-MRI). This estimation requires both high temporal and spatial resolution to capture the rapid tracer kinetic and detect small impairments and reliably discriminate boundaries. With this in mind, we propose a compressed sensing approach to decrease the acquisition time without sacrificing the reconstruction, especially in the region affected by the tracer. Within the framework of a TV-L1-L2 minimization for solving the reconstruction from partial Fourier data, we introduce a novel baseline-constraining term weighting the difference of the reconstructed volume from the baseline in all regions where no perfusion is apparent. We show that the proposed reconstruction scheme is able to provide accurate estimation of the tracer kinetics (the necessary step for estimating CBF, CBV and MTT) in the volume even at high acceleration (x16), with a RMSE of 11, a third of what achievable without the baseline constraint.