Reblurred deconvolution method for chemical shift removal in F‐19 (PFOB) MR imaging

Perfluorocarbons such as perfluoroctylbromide (PFOB) can be used as contrast agents in the vascular system for fluorine‐19 magnetic resonance imaging or as synthetic oxygen carriers. F‐19 imaging has been proposed for studying the vascular system, capillary flow, tissue perfusion, and tumor oxygenation. A major difficulty is that F‐19 compounds often have complex multipeak spectra. These peaks result in chemical shift artifacts, lower signal‐to‐noise ratios, and blurred images. Each peak also excites a different section when a section‐select gradient is applied. Direct inverse filtering is the simplest deconvo‐lution method for correcting such artifacts; however, two major difficulties present themselves: functional singularity and noise amplification at high frequencies. The use of a new reblurred de‐convolution (RED) method appears to overcome these problems. Although this method is based on iterative deconvolution in the spatial domain, the computational overhead is negligible. Since the point spread function and object data are already available in the time domain as FID data, RED appears to be useful for eliminating chemical shift artifacts and suppressing noise amplification while restoring the original image without loss of resolution.