View in EDS

A CVAE-based generative model for generalized B1 inhomogeneity corrected chemical exchange saturation transfer MRI at 5 T

Saved in:
Bibliographic Details
Title: A CVAE-based generative model for generalized B1 inhomogeneity corrected chemical exchange saturation transfer MRI at 5 T
Authors: Ruifen Zhang, Qiyang Zhang, Yin Wu
Source: NeuroImage, Vol 312, Iss, Pp 121202-(2025)
Publisher Information: Elsevier BV, 2025.
Publication Year: 2025
Subject Terms: Conditional variational autoencoder (CVAE), Adult, Deep Learning, Chemical exchange saturation transfer (CEST), Image Processing, Computer-Assisted, B1 inhomogeneity, Humans, Brain, Neurosciences. Biological psychiatry. Neuropsychiatry, Magnetic Resonance Imaging, RC321-571
Description: 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 B1 level. Spatial inhomogeneity of B1 field would bias CEST measurement. Conventional interpolation-based B1 correction method required CEST dataset acquisition under multiple B1 levels, substantially prolonging scan time. The recently proposed supervised deep learning approach reconstructed B1 inhomogeneity corrected CEST effect at the identical B1 as of the training data, hindering its generalization to other B1 levels. In this study, we proposed a Conditional Variational Autoencoder (CVAE)-based generative model to generate B1 inhomogeneity corrected Z spectra from single CEST acquisition. The model was trained from pixel-wise source-target paired Z spectra under multiple B1 with target B1 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 B1 inhomogeneity corrected CEST MRI. Results showed that the generated B1-corrected Z spectra agreed well with the reference averaged from regions with subtle B1 inhomogeneity. Moreover, the performance of the proposed model in correcting B1 inhomogeneity in APT CEST effect, as measured by both MTRasym and [Formula: see text] at 3.5 ppm, were superior over conventional Z/contrast-B1-interpolation and other deep learning methods, especially when target B1 were not included in sampling or training dataset. In summary, the proposed model allows generalized B1 inhomogeneity correction, benefiting quantitative CEST MRI in clinical routines.
Document Type: Article
Language: English
ISSN: 1095-9572
DOI: 10.1016/j.neuroimage.2025.121202
Access URL: https://pubmed.ncbi.nlm.nih.gov/40268259
https://doaj.org/article/5020824dce994853bd9304a5c9daf2ba
Accession Number: edsair.pmid.dedup....04a91875b02bfa1d717d6ca5f90409c9
Database: OpenAIRE
Description
Abstract: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 B1 level. Spatial inhomogeneity of B1 field would bias CEST measurement. Conventional interpolation-based B1 correction method required CEST dataset acquisition under multiple B1 levels, substantially prolonging scan time. The recently proposed supervised deep learning approach reconstructed B1 inhomogeneity corrected CEST effect at the identical B1 as of the training data, hindering its generalization to other B1 levels. In this study, we proposed a Conditional Variational Autoencoder (CVAE)-based generative model to generate B1 inhomogeneity corrected Z spectra from single CEST acquisition. The model was trained from pixel-wise source-target paired Z spectra under multiple B1 with target B1 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 B1 inhomogeneity corrected CEST MRI. Results showed that the generated B1-corrected Z spectra agreed well with the reference averaged from regions with subtle B1 inhomogeneity. Moreover, the performance of the proposed model in correcting B1 inhomogeneity in APT CEST effect, as measured by both MTRasym and [Formula: see text] at 3.5 ppm, were superior over conventional Z/contrast-B1-interpolation and other deep learning methods, especially when target B1 were not included in sampling or training dataset. In summary, the proposed model allows generalized B1 inhomogeneity correction, benefiting quantitative CEST MRI in clinical routines.
ISSN:10959572
DOI:10.1016/j.neuroimage.2025.121202