PEAKIT: A Gaussian Process regression analysis tool for chemical exchange saturation transfer spectra

Abstract

Chemical Exchange Saturation Transfer (CEST) is a powerful technique for metabolic imaging, capable of exploring concentrations in the μM to mM range. However, extracting quantitative information from Z-spectra can be challenging due to the non-CEST contributions present and the limited knowledge about the exchanging pools. The PEAKIT tool is proposed as an alternative approach to quantifying CEST peaks, which requires no prior assumptions about the frequency offset or the underlying shape of the baseline. Specifically, the tool takes as input an experimental Z-spectrum and proceeds to identify peak candidates. After a baseline estimation based on Gaussian Process regression, PEAKIT outputs the chemical shift offsets, the areas, the heights and the statistical significance of the detected peaks. The performance and limitations of the PEAKIT tool are discussed for in vitro and in vivo applications.

Keywords: Chemical Exchange Saturation Transfer (CEST); Gaussian process regression; Noise level; Peak detection; Python; Tkinter; User interface; Z-spectrum.