Feasibility of Singular Value Decomposition Implementation for 129 Xe Hyperpolarized Chemical Exchange Saturation Transfer ( HyperCEST ) Molecular Imaging

ABSTRACT Purpose To develop a method for enhanced quantification of hyperpolarized chemical exchange saturation transfer (HyperCEST) using singular value decomposition (SVD) for both spectroscopy and molecular imaging. Methods Dynamic HyperCEST spectra acquired with cucurbit[6]uril (CB[6]) and resorcinarene trimer methanesulfonate (R3‐Noria‐MeSO 3 H) in aqueous solution were subjected to SVD to denoise the magnetic resonance spectra. The HyperCEST depletion spectra were recalculated from SVD‐reconstructed data using a standard approach. The on‐ and off‐resonance coronal 129 Xe images acquired from R3‐Noria‐MeSO 3 H samples in vitro and from three healthy rats post CB[6] injection were also reconstructed using SVD prior to pixel‐by‐pixel HyperCEST map calculation. A localized variance of the HyperCEST values was calculated. Results A paired t ‐test demonstrated a statistically significant increase (up to ∼13%) in localized HyperCEST values both in vivo (in the animal kidneys and aorta) and in vitro (across the phantom volume) due to the SVD incorporation. In addition, SVD significantly reduced HyperCEST variance. The HyperCEST variance for in vivo images was reduced by 32.3%, whereas the variance for in vitro images was reduced by 36%. The SNR of the SVD‐reconstructed images and spectra was significantly higher. The SVD resulted in a significant improvement of the shape of HyperCEST depletion spectra and maximized the HyperCEST effect in the case of CB[6] for lower depolarization angles. Conclusion Implementation of the proposed SVD‐based method is beneficial for 129 Xe HyperCEST molecular imaging due to significantly improved quantification of the HyperCEST effect, reduced variability of the localized HyperCEST values, and significantly improved quality of the HyperCEST depletion spectra.