Andre Fischer1,2, Thomas Christian Basse-Lsebrink2,3, Thomas Kampf2,4, Gesa Ladewig3, Martin Blaimer1, Felix Breuer1, Guido Stoll3, Wolfgang Rudolf Bauer4, Peter Jakob1,2
1Research Center Magnetic Resonance Bavaria e.V., Wrzburg, Germany; 2Department for Experimental Physics 5, University of Wrzburg, Wrzburg, Germany; 3Neurology, University of Wrzburg, Wrzburg, Germany; 4Medical Clinic and Polyclinic I, University of Wrzburg, Wrzburg, Germany
19F imaging suffers from low signal intensities and long data acquisition times due to averaging. Since the distribution of the 19F signal is sparse in the image domain, we propose a Compressed Sensing (CS) reconstruction schema. CS allows the reconstruction of sparse undersampled datasets. Classical Chemical Shift Imaging (CSI) is a purely phase encoded technique; therefore, random sampling can be applied which is optimal for the CS reconstruction due to the incoherent undersampling artifacts. In this work we demonstrate the potential of Nonconvex CS  in cellular imaging using 19F CSI. This study is the first one where CS exploits solely the sparsity in image space and not in the spectral dimension. Initial results from retrospectively undersampled phantom and in-vivo experiments are presented. These experiments demonstrate that it is possible to obtain almost the same information content by collecting only a significantly reduced fraction of all phase encoding steps in 19F CSI.