Quantitative Chemical Exchange Saturation Transfer Imaging with Golden-Angle Radial k-Space and Locally Low-Rank Reconstruction

Abstract: Purpose: To develop a motion-robust and geometrically-accurate quantitative CEST approach using radial k-space sampling, locally low-rank reconstruction and neural network quantification. Methods: The acquisition schedule was generated via deep learning optimization. The optimized sequence accuracy was validated in numerical simulations in digital phantoms. The spokes per measurement count was optimized using simulations and in vivo ablation studies in a healthy subject. Five healthy subjects were repeatedly scanned, and regions of interest were defined. Tissue maps from the proposed sequence were compared to an EPI-based quantitative CEST sequence. Motion sensitivity and test-retest reproducibility was assessed using the coefficient of variation (CV) and intraclass-correlation coefficient (ICC). Results: 3D quantitative CEST maps were acquired in 11 minutes using 34 spokes per measurement. Numerical simulations showed a mean error of <14% for all tissue parameters. In vivo tissue-parameter values agreed well with prior CEST-MRF studies in brain. The mean ICC over all tissue maps was 0.92 in white matter (WM) and 0.87 in grey matter (GM). The mean inter-subject CV was 5.4%/3.4% (WM/GM) and the motion vs. no-motion error was 265% for EPI but 8.6% for the radial acquisition. Conclusion: A motion-robust and geometrically accurate quantitative radial CEST pulse sequence and reconstruction framework is demonstrated. This approach enables accurate, reproducible 3D brain quantitative CEST imaging in clinically relevant scan times.