Quantification of the CEST effect by Gaussian mixture modeling of Z-spectrum

Quantitative evaluation of chemical exchange saturation transfer (CEST) is usually done by solving Bloch-McConnell equations (BME). BMEs are not easily extended and applying them to describe the multi-pool data involves a complex process. In this paper, we developed a Gaussian mixture model (GMM) to represent each component involved in the Z-spectrum by a Gaussian distribution. We then tested and evaluated the GMM for the two-pool exchange site and experimental data. The results showed that GMM is able to fit the experimental data and its accuracy is almost similar to that of the BME model. (average percent of Relative Sum Square Error (%RSSE) <;0.6). Accuracy and simplicity were found to be the advantages of the GMM and lack of analytical relationships among the GMM parameters and physical characteristics of the CEST effect turned out to be its main limitations. We quantified contrast agent (CA) concentration (population fraction of CEST pool) and chemical exchange rate applying the GMM to the simulated data of a two-pool exchange site. It was found that the means and variances of the Gaussians can be used for this purpose. In addition, GMM determines the resonance frequency of each pool easily and accurately because these frequencies are equal to the mean values of GMM.