Answers are grounded in this paper and cite their sources.
This study proposes a deep learning-based method termed frequency-flexible chemical exchange saturation transfer (CEST) imaging network (FlexCENT), which enables robust CEST quantification across variable frequency offset schemes without requiring retraining. FlexCENT integrates frequency offset encoding with a three-dimensional (3D) U-Net to process CEST images and frequency offsets as inputs and predict Lorentzian parameters of the 4-pool model (water, MT, APT, rNOE), including B 0 inhomogeneity. By transforming frequency offsets into a continuous spectral feature representation, the frequency offset encoding allows FlexCENT to generalize to unseen frequency offset schemes. Trained on synthetic data generated from the 4-pool Lorentzian model, FlexCENT was validated through numerical simulations, tumor-bearing mouse experiments, and a human brain experiment, alongside comparisons with 4-pool Lorentzian fitting, DeepCEST, and LKAN networks. The results demonstrate that FlexCENT successfully quantified CEST parameters across all experiments, maintaining consistent performance under varying frequency offset conditions without retraining. It exhibited superior noise robustness in numerical simulations and enhanced anatomical delineation in vivo parametric mapping compared to other methods. In conclusion, by combining spectral information with spatial information, FlexCENT provides an efficient, flexible, and robust quantitative approach for CEST imaging. It significantly enhance the quantification capability and clinical potential of CEST imaging.
No takes yet. Share an insight, caveat, or question.
Yu et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: