Deep learning water-unsuppressed MRSI at ultra-high field for simultaneous quantitative metabolic, susceptibility and myelin water imaging

2025-12-16

AI Summary

Water sideband contamination in 7T unsuppressed magnetic resonance spectroscopic imaging (MRSI) impedes robust metabolite quantification. Method: We propose ECCENTRIC non-Cartesian sampling combined with an ultra-short echo time (TE) sequence, enabling simultaneous acquisition of metabolite maps, quantitative susceptibility mapping (QSM), and myelin water fraction (MWF) within a single 12-minute scan. We introduce WALINET+, a novel physics-informed deep learning network that jointly performs spectral fitting and artifact suppression. Contribution/Results: WALINET+ enables robust removal of water, lipid, and sideband signals—eliminating the need for water suppression or separate water reference scans—and supports absolute, three-modal quantitative integration. Simulations yield NRMSE < 2%; in vivo metabolite quantification error is < 20% (limits of agreement [LoA] ±63%). QSM and MWF show mean deviations of 0 ppm and 5.5% versus GRE-based gold standards (LoA ±0.05 ppm and ±12.75%, respectively). SNR of creatine, glutamate, and other metabolites is significantly enhanced.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Multi-modal Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Web data integration and cleaning

Abstract

Purpose: Magnetic Resonance Spectroscopic Imaging (MRSI) maps endogenous brain metabolism while suppressing the overwhelming water signal. Water-unsuppressed MRSI (wu-MRSI) allows simultaneous imaging of water and metabolites, but large water sidebands cause challenges for metabolic fitting. We developed an end-to-end deep-learning pipeline to overcome these challenges at ultra-high field. Methods:Fast high-resolution wu-MRSI was acquired at 7T with non-cartesian ECCENTRIC sampling and ultra-short echo time. A water and lipid removal network (WALINET+) was developed to remove lipids, water signal, and sidebands. MRSI reconstruction was performed by DeepER and a physics-informed network for metabolite fitting. Water signal was used for absolute metabolite quantification, quantitative susceptibility mapping (QSM), and myelin water fraction imaging (MWF). Results: WALINET+ provided the lowest NRMSE (< 2%) in simulations and in vivo the smallest bias (< 20%) and limits-of-agreement (+-63%) between wu-MRSI and ws-MRSI scans. Several metabolites such as creatine and glutamate showed higher SNR in wu-MRSI. QSM and MWF obtained from wu-MRSI and GRE showed good agreement with 0 ppm/5.5% bias and +-0.05 ppm/ +- 12.75% limits-of-agreement. Conclusion: High-quality metabolic, QSM, and MWF mapping of the human brain can be obtained simultaneously by ECCENTRIC wu-MRSI at 7T with 2 mm isotropic resolution in 12 min. WALINET+ robustly removes water sidebands while preserving metabolite signal, eliminating the need for water suppression and separate water acquisitions.

Problem

• Develops deep learning to remove water sidebands in unsuppressed MRSI
• Enables simultaneous metabolic, susceptibility, and myelin water imaging
• Achieves high-resolution quantitative brain mapping at ultra-high field

Innovation

• Deep learning pipeline for water-unsuppressed MRSI at 7T
• WALINET+ network removes water, lipids, and sidebands
• Physics-informed network enables simultaneous metabolic and water imaging