Application of a Markov chain Monte Carlo method for robust quantification in chemical exchange saturation transfer magnetic resonance imaging.

BACKGROUND

Chemical exchange saturation transfer (CEST) magnetic resonance imaging can provide surrogate biomarkers for disease diagnosis. However, endogenous CEST effects are always diluted and contaminated by competing effects, which results in unwanted signal contributions that lessen the specificity of CEST to underlying biochemical exchange processes. The aim of this study was to examine a method for the accurate quantification of CEST effects.

METHODS

A Markov chain Monte Carlo (MCMC)-based Bayesian inference approach was proposed to estimate the exchange parameters, and CEST effects could be fitted using these estimations. This approach was tested in Bloch simulation and ischemic stroke rat experiments, and its performance was evaluated using quantification maps and numerical metrics.

RESULTS

With 12 groups of simulations, the MCMC method achieved satisfactory fittings on both 2-pool and 5-pool models. The sum of squares error values and the root mean square error of the fitted Z-spectra were smaller than 10, and the coefficient of determination (R-squared) values were close to 1. The corresponding CEST quantification spectra were also well fitted and successfully separated the mixed CEST effects. The estimated parameters showed little bias relative to the ground truth, with errors between the true and estimated values of each parameter of less than 0.5%. In the animal experiments, fitted using the MCMC method showed obvious contrast between ischemic and contralateral regions at the early stage. Compared with other quantification methods, it displayed the highest contrast-to-noise ratios (3.9, 2.73, and 3.93) and the lowest coefficient of variation values (0.181, 0.2224, and 0.2897) in all three stroke periods.

CONCLUSIONS

The MCMC method provided an efficient approach for parameter estimation and CEST effect quantification. The method may therefore be useful in achieving an accurate pathological diagnosis.