A CVAE-based generative model for generalized B 1 inhomogeneity corrected chemical exchange saturation transfer MRI at 5 T

作者：Ruifen Zhang, Qiyang Zhang, Yin Wu · 发表于：NeuroImage · 年份：2025 · DOI：10.1016/j.neuroimage.2025.121202 · 被引用次数：5 · 研究领域：Lanthanide and Transition Metal Complexes、Advanced MRI Techniques and Applications、MRI in cancer diagnosis

Chemical exchange saturation transfer (CEST) magnetic resonance imaging (MRI) has emerged as a powerful tool to image endogenous or exogenous macromolecules. CEST contrast highly depends on radiofrequency irradiation B 1 level. Spatial inhomogeneity of B 1 field would bias CEST measurement. Conventional interpolation-based B 1 correction method required CEST dataset acquisition under multiple B 1 levels, substantially prolonging scan time. The recently proposed supervised deep learning approach reconstructed B 1 inhomogeneity corrected CEST effect at the identical B 1 as of the training data, hindering its generalization to other B 1 levels. In this study, we proposed a Conditional Variational Autoencoder (CVAE)-based generative model to generate B 1 inhomogeneity corrected Z spectra from single CEST acquisition. The model was trained from pixel-wise source–target paired Z spectra under multiple B 1 with target B 1 as a conditional variable. Numerical simulation and healthy human brain imaging at 5 T were respectively performed to evaluate the performance of proposed model in B 1 inhomogeneity corrected CEST MRI . Results showed that the generated B 1 -corrected Z spectra agreed well with the reference averaged from regions with subtle B 1 inhomogeneity. Moreover, the performance of the proposed model in correcting B 1 inhomogeneity in APT CEST effect, as measured by both M T R a s y m and M T R R e x at 3.5 ppm, were superior over conventional Z/contrast- B 1 -interpolation ...