Acceleration of conventional data acquisition in dynamic contrast enhancement: Comparing keyhole approaches with compressive sensing

Dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) has become a valuable clinical tool for cancer diagnosis and prognosis. DCE MRI provides pharmacokinetic parameters dependent on the extravasation of small molecular contrast agents, and thus high temporal resolution and/or spatial resolution is required for accurate estimation of parameters. In this article we investigate the efficacy of 2 under sampling approaches to speed up DCE MRI: a conventional keyhole approach and compressed sensing–based imaging. Data reconstructed from variants of these methods has been compared with the full k-space reconstruction with respect to data quality and pharmacokinetic parameters K^trans and v_e. Overall, compressive sensing provides better data quality and reproducible parametric maps than key-hole methods with higher acceleration factors. In particular, an under sampling mask based on a priori precontrast data showed high fidelity of reconstructed data and parametric maps up to 5× acceleration.