Development and validation of a smoothing-splines-based correction method for improving the analysis of CEST-MR images.

Chemical exchange saturation transfer (CEST) imaging is an emerging MRI technique relying on the use of endogenous or exogenous molecules containing exchangeable proton pools. The heterogeneity of the water resonance frequency offset plays a key role in the occurrence of artifacts in CEST-MR images. To limit this drawback, a new smoothing-splines-based method for fitting and correcting Z-spectra in order to compensate for low signal-to-noise ratio (SNR) without any a priori model was developed. Global and local voxel-by-voxel Z-spectra were interpolated by smoothing splines with smoothing terms aimed at suppressing noise. Thus, a map of the water frequency offset ('zero' map) was used to correctly calculate the saturation transfer (ST) for each voxel. Simulations were performed to compare the method to polynomials and zero-only-corrected splines on the basis of SNR improvement. In vitro acquisitions of capillaries containing solutions of LIPOCEST agents at different concentrations were performed to experimentally validate the results from simulations. Additionally, ex vivo investigations of bovine muscle mass injected with LIPOCEST agents were performed as a function of increasing pulse power. The results from simulations and experiments highlighted the importance of a proper 'zero' correction (15% decrease of fictitious CEST signal in phantoms and ex vivo preparations) and proved the method to be more accurate compared with the previously published ones, often providing a SNR higher than 5 in different simulated and experimentally noisy conditions. In conclusion, the proposed method offers an accurate tool in CEST investigation. Copyright (c) 2008 John Wiley & Sons, Ltd.