Deuterium imaging of the Warburg effect at sub‐millimolar concentrations by joint processing of the kinetic and spectral dimensions

Abstract

Deuterium metabolic imaging (DMI) is a promising molecular MRI approach, which follows the administration of deuterated substrates and their metabolization. [6,6’‐2H2]‐glucose for instance is preferentially converted in tumors to [3,3’‐2H2]‐lactate as a result of the Warburg effect, providing a distinct resonance whose mapping using time‐resolved spectroscopic imaging can diagnose cancer. The MR detection of low‐concentration metabolites such as lactate, however, is challenging. It has been recently shown that multi‐echo balanced steady‐state free precession (ME‐bSSFP) increases the signal‐to‐noise ratio (SNR) of these experiments approximately threefold over regular chemical shift imaging; the present study examines how DMI's sensitivity can be increased further by advanced processing methods. Some of these, such as compressed sensing multiplicative denoising and block‐matching/3D filtering, can be applied to any spectroscopic/imaging methods. Sensitivity‐enhancing approaches were also specifically tailored to ME‐bSSFP DMI, by relying on priors related to the resonances' positions and to features of the metabolic kinetics. Two new methods are thus proposed that use these constraints for enhancing the sensitivity of both the spectral images and the metabolic kinetics. The ability of these methods to improve DMI is evidenced in pancreatic cancer studies carried at 15.2 T, where suitable implementations of the proposals imparted eightfold or more SNR improvement over the original ME‐bSSFP data, at no informational cost. Comparisons with other propositions in the literature are briefly discussed.

Whole‐brain deuterium metabolic imaging via concentric ring trajectory readout enables assessment of regional variations in neuronal glucose metabolism

“…Improved precision and accuracy to predict metabolite concentrations compared with the simulated gold standard has been shown after low-rank denoising of synthetic data with a similar noise level compared with in vivo datasets. The fact, that averaged time courses of the gold standard presented from synthetic data (featuring small standard deviations; see Figure S4) is caused by the partial volume effects Other techniques such as multiecho steady state free precession DMI (Montrazi et al, 2023;Peters et al, 2021) or deep learning approaches (Li et al, 2021) could also significantly increase the achievable SNR and consequently the spatial resolution.…”

High-sensitivity deuterium metabolic MRI differentiates acute pancreatitis from pancreatic cancers in murine models

“…The rst four points of this FID had to be discarded due to the digital ltering. Chemical shift images were then obtained from the resulting FID using the IDEAL tting method aided by 1 H-based eld mappings; note that additional re nements incorporating ts along the kinetic dimension and denoising techniques, [19] could also be directly applied to these CSI-SSFP acquisitions.…”

“…Spatial domains for both the ME-SSFP and CSI-SSFP experiments were reconstructed identically, by 2D Fourier transform after zero-lling to 64x64 points. Images arising from the separated ME-SSFP echoes and for each CSI-SSFP FID point (5 frames for ME-SSFP, and 42 for CSI-SSFP from 46 acquired points) were processed using the RK-SpecRecon (Regularized kinetics spectral reconstruction) algorithm, [19] which isolated the images of the individual sites using a priori known chemical shift positions (4.7, 3.6 and 1.2 ppm for the three DMI metabolites; 2 ppm carrier frequency) and regularization of the metabolites' evolution along the kinetic series. 1 H-based B 0 maps were used as initial guesses in the tting to avoid "swaps" otherwise observed upon processing.…”

“…[18,21] Acquiring ve echoes for separating these three species gives the least-square tting protocol used to perform the data analysis, [23,24] with extra robustness and an ability to incorporate eld inhomogeneities; this was enhanced by the use of periodically collected 1 H-based abdominal eld maps. [19] The ME-SSFP sequence also controlled the readout bandwidth, voxel size, and eld of views (FOV) of the in-plane imaging domains; the remaining dimension was accounted for by a slice-selective pulse. Since processing ME-SSFP's odd and even echoes jointly introduced data inconsistencies, y-back gradients were thus used, and only odd echoes were processed (Fig.…”

“…These tests were initially conducted using multi-echo balanced SSFP (ME-SSFP) approaches, for which recent ndings demonstrated ca 3-8 times signal-to-noise ratio (SNR) improvements over DMI data collected by conventional chemical shift imaging (CSI) techniques. [18,19] While having no problems in imaging sub-mM [1-13 C 1 ]-lactate concentration levels in tumor models, these DMI experiments failed to detect lactate in mice subject to chemically-induced acute pancreatitis. To enhance sensitivity to lactate production even further, we explored the utilization of new SSFP variant, based on a weighted average k-space acquisition of CSI-SSFP data.…”