Double-step R1rho-based LOrentzian Fitting (DROF): a new CEST analysis approach and its comparison with existing methods

Motivation: At 3T MRI, reliable quantification of amide, guan, and rNOE is hindered by broad background and fitting dependency on initial values and boundary conditions. Goal(s): To propose a robust CEST analysis method for extracting 5-pool parameters at 3T and a quantitative metric to evaluate the performance of existing fitting methods. Approach: A double-step R1rho-based LOrentzian Fitting (DROF) method with random initial guesses was developed for CEST analysis. The accuracy and robustness of DROF were assessed on both simulation and in-vivo human data at 3T and compared with existing methods. Results: The new method demonstrates improved specificity and stability compared to other methods. Impact: We proposed a double-step R1rho-based LOrentzian Fitting (DROF) method for extracting 5-pool CEST parameters at 3T. DROF exhibits enhanced specificity and stability compared to other methods, suggesting its potential to promote wider CEST applications at 3T.