Performance of compressed sensing for fluorine‐19 magnetic resonance imaging at low signal‐to‐noise ratio conditions

Abstract

Purpose:To examine the performance of compressed sensing (CS) in reconstructing low signal-to-noise ratio (SNR) 19 F MR signals that are close to the detection threshold and originate from small signal sources with no a priori known location. Methods: Regularization strength was adjusted automatically based on noise level.As performance metrics, root-mean-square deviations, true positive rates (TPRs), and false discovery rates were computed. CS and conventional reconstructions were compared at equal measurement time and evaluated in relation to high-SNR reference data. 19 F MR data were generated from a purpose-built phantom and benchmarked against simulations, as well as from the experimental autoimmune encephalomyelitis mouse model. We quantified the signal intensity bias and introduced an intensity calibration for in vivo data using high-SNR ex vivo data. Results: Low-SNR 19 F MR data could be reliably reconstructed. Detection sensitivity was consistently improved and data fidelity was preserved for undersampling and averaging factors of α = 2 or = 3. Higher α led to signal blurring in the mouse model.

First in vivo fluorine-19 magnetic resonance imaging of the multiple sclerosis drug siponimod

“…Multiple publications have demonstrated that the increased averaging enabled by CS can lead to sensitivity improvements for low SNR MRI [86,115,116]. The distribution of siponimod is less sparse than that of inflammatory lesions [86]. At the spatial resolution achieved in this study, the signal distribution has important features at the scale of single voxels, and thus does not possess the sparsity necessary for successful application of conventional CS techniques.…”

“…Similarly shortened relaxation times were observed in ex vivo tissue, and we expect that the properties in in vivo tissues will not differ substantially. The effect of protein binding on the transverse relaxation times (reduction by up to 98%) creates challenges for pulse sequences which are employed in the more common 19 F MR methods such as point resolved spectroscopy (PRESS) or highly accelerated RARE in the case of perfluorocarbon imaging [37,[84][85][86].…”

“…The encoding trajectories commonly employed in UTE sequences are well-suited for undersampling and compressed sensing (CS) image reconstruction [112][113][114]. Multiple publications have demonstrated that the increased averaging enabled by CS can lead to sensitivity improvements for low SNR MRI [86,115,116]. The distribution of siponimod is less sparse than that of inflammatory lesions [86].…”

Single Fluorinated Agent for Multiplexed 19F‐MRI with Micromolar Detectability Based on Dynamic Exchange

“…Several strategies were proposed to enhance 19 F‐MRI sensitivity including the use of high‐end cryogenic rf coils, [20] paramagnetic relaxation enhancement, [21] crystallographic defect induction, [22] compressed‐sensing‐based schemes [23] and even hyperpolarization techniques [24] . The 19 F‐GEST approach shown here offers an alternative strategy to detect very low levels of thermally polarized pools of fluorine‐19 spins with readily available hardware and can open new avenues for the development of advanced molecular architectures for the growing field of hotspot 19 F‐MRI, which, so far, has not been used for mapping targets at very low concentrations.…”

Data Preparation Protocol for Low Signal-to-Noise Ratio Fluorine-19 MRI

“…2d). The shown example is based on a dataset with fluorine nanoparticle labeled immune cells showing the sites of inflammation in a mouse model of multiple sclerosis (see Setup 4 in Starke et al [12] for details). The data preparation protocol is equally applicable to low SNR 19 F MRI of the kidney.…”