Shihao Zeng

CEST analysis using a deep residual learning method based on Bloch–McConnell synthetic training data

S. Zeng, H. Zhang, Z. Wang, J. Wang, P. Cai, J. Huang

ISMRM & ISMRT Annual Meeting · 2025 · first author

Abstract

Conventional Lorentzian fitting for CEST MRI is time-consuming and sensitive to initial values and bounds. This work uses a deep residual network, trained only on Bloch–McConnell synthetic data, to predict the reference Z-spectrum and quantify CEST contrast. On simulations the estimates agree closely with ground truth, and the method is feasible for in vivo human brain data from multiple sclerosis and healthy controls.