Yitian Fan, Chongxue Bie, Xinding Jiang, Zhekai Chen, Meijuan Li, Yi Zhou, Lin Chen
PURPOSE: To develop a motion-resolved acquisition and reconstruction framework for motion-robust and spectrally reliable abdominal CEST imaging under free-breathing conditions. THEORY AND METHODS: A framework termed Golden-angle RAdial CEST MRI with MOtion-REsolved reconstruction (GRACE-MORE) was developed to improve acquisition, respiratory-phase binning, and reconstruction. An interleaved steady-state saturation module with golden-angle radial sampling was adopted to enhance saturation efficiency and ensure uniform k-space coverage. Respiratory-phase binning was performed using a self-gated strategy, in which respiratory amplitude and phase were extracted from central k-space via principal component analysis, followed by hybrid empirical mode decomposition-continuous wavelet transform sorting to correct baseline drift and irregular breathing. Reconstruction was performed using a deep unrolled network constrained by low-rank and sparsity priors, which incorporated a modified U-Net with channel attention and a sliding-window grouping scheme to capture spatio-spectral correlations across saturation offsets. RESULTS: Validation on simulated, preclinical, and clinical datasets demonstrated more than a threefold improvement in binning accuracy compared with conventional methods. GRACE-MORE reduced reconstruction error, yielded a higher structural similarity index and peak signal-to-noise ratio in simulations, and achieved improved anatomical fidelity, effective motion suppression, and up to a fourfold increase in Z-spectral signal-to-noise ratio in vivo. CONCLUSION: GRACE-MORE enables motion-robust and spectrally reliable abdominal CEST imaging under free-breathing conditions.