Abstract
Chemical exchange saturation transfer (CEST) MRI detects exchangeable protons in metabolites, with signal governed by exchange rate (k­ex), T2, and pool size (f). Reproducibility is hindered by k­ex discrepancies across studies, arising from fitting parameter coupling, buffer conditions, experimental design (e.g., QUEST, QUESP, QUASS), and choice of analysis régime (steady-state vs. non-steady-state; SS vs. NSS).Using ammonium chloride (NH4Cl) as a model analyte, we investigated how buffering conditions, experimental design, and fitting model affect k­ex estimation. Multi-concentration CEST experiments were performed at 7T across saturation times and powers, comparing citrate buffer, common in the literature, to potassium hydrogen phthalate (KHP) as an amphoteric monoprotic alternative. Confidence intervals were assessed across buffer conditions and fitting models (single- vs. multi-concentration; SS vs. NSS). Parameter coupling was characterised via Markov chain Monte Carlo (MCMC) sampling.Findings highlight significant differences in kex under unbuffered (260 ± 4 Hz), KHP-buffered (180 ± 10 Hz), and citrate-buffered (160 ± 7 Hz) conditions. Multi-concentration NSS fitting effectively decoupled kex, f, and T2 (unbuffered fitting confidence intervals: 4 Hz, 0.04%, 10 ms, respectively), demonstrating robustness over single-concentration (3 Hz, 0.02%, 10600 ms) and SS (8 Hz, 0.09%, 190 ms) fits.These findings emphasize the importance of experimental design and analysis in improving the reliability of kex measurements, providing a pathway for more interpretable CEST experiments. Multi-concentration NSS fitting decouples CEST parameters, and buffer selection should reflect both the ionic strength and buffering capacity of the target biological compartment for exchange rates to be meaningfully comparable across studies.

This publication has 18 references indexed in Scilit: