Comparison of model‐free Lorentzian and spinlock model‐based fittings in quantitative CEST imaging of acute stroke

Abstract

PurposeCEST MRI detects complex tissue changes following acute stroke. Our study aimed to test if spinlock model‐based fitting of the quasi‐steady‐state (QUASS)‐reconstructed equilibrium CEST MRI improves the determination of multi‐pool signal changes over the commonly‐used model‐free Lorentzian fitting in acute stroke.Theory and MethodsMultiple three‐pool CEST Z‐spectra were simulated using Bloch‐McConnell equations for a range of T1, relaxation delay, and saturation times. The multi‐pool CEST signals were solved from the simulated Z‐spectra to test the accuracy of routine Lorentzian (model‐free) and spinlock (model‐based) fittings without and with QUASS reconstruction. In addition, multiparametric MRI scans were obtained in rat models of acute stroke, including relaxation, diffusion, and CEST Z‐spectrum. Finally, we compared model‐free and model‐based per‐pixel CEST quantification in vivo.ResultsThe spinlock model‐based fitting of QUASS CEST MRI provided a nearly T1‐independent determination of multi‐pool CEST signals, advantageous over the fittings of apparent CEST MRI (model‐free and model‐based). In vivo data also demonstrated that the spinlock model‐based QUASS fitting captured significantly different changes in semisolid magnetization transfer (−0.9 ± 0.8 vs. 0.3 ± 0.8%), amide (−1.1 ± 0.4 vs. −0.5 ± 0.2%), and guanidyl (1.0 ± 0.4 vs. 0.7 ± 0.3%) signals over the model‐free Lorentzian analysis.ConclusionOur study demonstrated that spinlock model‐based fitting of QUASS CEST MRI improved the determination of the underlying tissue changes following acute stroke, promising further clinical translation of quantitative CEST imaging.