Assessment of Sodium MRI at 7 Tesla as Predictor of Therapy Response and Survival in Glioblastoma Patients.

The purpose of this work was to prospectively investigate sodium (23Na) MRI at 7 Tesla (T) as predictor of therapy response and survival in patients with glioblastoma (GBM). Thus, 20 GBM patients underwent 23Na MRI at 7T before, immediately after and 6 weeks after chemoradiotherapy (CRT). The median tissue sodium concentration (TSC) inside the whole tumor excluding necrosis was determined. Initial response to CRT was assessed employing the updated response assessment in neuro-oncology working group (RANO) criteria. Clinical parameters, baseline TSC and longitudinal TSC differences were compared between patients with initial progressive disease (PD) and patients with initial stable disease (SD) using Fisher's exact tests and Mann-Whitney-U-tests. Univariate proportional hazard models for progression free survival (PFS) and overall survival (OS) were calculated using clinical parameters and TSC metrics as predictor variables. The analyses demonstrated that TSC developed heterogeneously over all patients following CRT. None of the TSC metrics differed significantly between cases of initial SD and initial PD. Furthermore, TSC metrics did not yield a significant association with PFS or OS. Conversely, the initial response according to the RANO criteria could significantly predict PFS [univariate HR (95%CI) = 0.02 (0.0001-0.21), p < 0.001] and OS [univariate HR = 0.17 (0.04-0.65), p = 0.005]. In conclusion, TSC showed treatment-related changes in GBM following CRT, but did not significantly correlate with the initial response according to the RANO criteria, PFS or OS. In contrast, the initial response according to the RANO criteria was a significant predictor of PFS and OS. Future investigations need to elucidate the reasons for treatment-related changes in TSC and their clinical value for response prediction in glioblastoma patients receiving CRT.

Multiparametric MRI for Characterization of the Basal Ganglia and the Midbrain.

Objectives To characterize subcortical nuclei by multi-parametric quantitative magnetic resonance imaging. Materials and Methods: The following quantitative multiparametric MR data of five healthy volunteers were acquired on a 7T MRI system: 3D gradient echo (GRE) data for the calculation of quantitative susceptibility maps (QSM), GRE sequences with and without off-resonant magnetic transfer pulse for magnetization transfer ratio (MTR) calculation, a magnetization-prepared 2 rapid acquisition gradient echo sequence for T1 mapping, and (after a coil change) a density-adapted 3D radial pulse sequence for 23Na imaging. First, all data were co-registered to the GRE data, volumes of interest (VOIs) for 21 subcortical structures were drawn manually for each volunteer, and a combined voxel-wise analysis of the four MR contrasts (QSM, MTR, T1, 23Na) in each structure was conducted to assess the quantitative, MR value-based differentiability of structures. Second, a machine learning algorithm based on random forests was trained to automatically classify the groups of multi-parametric voxel values from each VOI according to their association to one of the 21 subcortical structures. Results The analysis of the integrated multimodal visualization of quantitative MR values in each structure yielded a successful classification among nuclei of the ascending reticular activation system (ARAS), the limbic system and the extrapyramidal system, while classification among (epi-)thalamic nuclei was less successful. The machine learning-based approach facilitated quantitative MR value-based structure classification especially in the group of extrapyramidal nuclei and reached an overall accuracy of 85% regarding all selected nuclei. Conclusion Multimodal quantitative MR enabled excellent differentiation of a wide spectrum of subcortical nuclei with reasonable accuracy and may thus enable sensitive detection of disease and nucleus-specific MR-based contrast alterations in the future.

Towards accelerated quantitative sodium MRI at 7 T in the skeletal muscle: Comparison of anisotropic acquisition- and compressed sensing techniques.

PURPOSE: To compare three anisotropic acquisition schemes and three compressed sensing (CS) approaches for accelerated tissue sodium concentration (TSC) quantification using 23Na MRI at 7 T. MATERIALS AND METHODS: Three anisotropic 3D-radial acquisition sequences were evaluated using simulations, phantom- and in vivo TSC measurements: An anisotropic density-adapted 3D-radial sequence (3DPR-C), a 3D acquisition-weighted density-adapted stack-of-stars sampling scheme (SOS) and a SOS approach with golden-ratio rotation (SOS-GR). Eight healthy volunteers were examined at a 7 Tesla MRI system. TSC measurements of the calf were conducted with a nominal spatial resolution of Δx = (3.0 × 3.0 × 15.0) mm3 and a field of view of (156.0 × 156.0 × 240.0) mm3 for multiple undersampling factors (USF). Three CS reconstructions were evaluated: Total variation CS (TV-CS), 3D dictionary-learning compressed sensing (3D-DLCS) and TV-CS with a block matching prior (TV-BL-CS). Results of the simulations and measurements were compared to a simulated ground truth (GT) or a fully sampled reference measurement (FS), respectively. The deviation of the mean TSC evaluated in multiple ROI (mEGT/FS) and the normalized root-mean-squared error (NRMSE) for simulations were evaluated for CS and NUFFT reconstructions. RESULTS: In simulations, the SOS-GR yielded the lowest NRMSE and mEGT (< 4%) with NUFFT for an acquisition time (TA) of less than 2 min. CS further improved the results. In simulations and measurements, the best TSC quantification results were obtained with 3D-DLCS and SOS-GR (lowest NRMSE, mEGT < 2.6% in simulations, mEGT < 10.7% for phantom measurements and mEFS < 6% in vivo) with an USF = 4.1 (TA < 2 min). TV-CS showed no or only slight improvements to NUFFT. The results of TV-BL-CS were similar to 3D-DLCS. DISCUSSION: The TA for TSC measurements could be reduced to less than 2 min by using adapted sequences such as SOS-GR and CS reconstruction approaches such as 3D-DLCS or TV-BL-CS, while the quantitative accuracy stays comparable to a fully sampled NUFFT reconstruction (approx. 8 min TA). In future, the lower TA could improve clinical applicability of TSC measurements.

Ultra-high-field sodium MRI as biomarker for tumor extent, grade and IDH mutation status in glioma patients.

PURPOSE: This prospective clinical trial investigated sodium (23Na) MRI at 7 Tesla (T) field strength as biomarker for tumor extent, isocitrate dehydrogenase (IDH) mutation and O6-methylguanine DNA methyltransferase (MGMT) promotor methylation in glioma patients. METHODS: 28 glioma patients underwent 23Na MRI on a 7T scanner (Siemens Healthcare, Erlangen, Germany) parallel to standard 3T MRI before chemoradiation. Areas of Gadolinium-contrast enhancement (gdce), non-enhancing T2-hyperintensity (regarded as edema), necrosis, and normal-appearing white matter (nawm) were segmented on 3T MRI imaging and were co-registered with the 23Na images. The median total 23Na concentrations of all areas were compared by pairwise t-tests. Furthermore, areas of gdce and edema were merged to yield the whole tumor area without necrosis. Subsequently, the difference in median of the 23Na concentration of this whole tumor area was compared between IDH-mutated and IDH wild-type gliomas as well as MGMT methylated and MGMT not-methylated glioblastomas using Whitney-Mann U-tests. All p-values were corrected after the Bonferroni-Holm procedure. RESULTS: The 23Na concentration increased successively from nawm to necrotic areas (mean ± sd: nawm = 37.84 ± 5.87 mM, edema = 54.69 ± 10.64 mM, gdce = 61.72 ± 12.95 mM, necrosis = 81.88 ± 17.53 mM) and the concentrations differed statistically significantly between all regarded areas (adjusted p-values for all pairwise comparisons < 0.05). Furthermore, IDH-mutated gliomas showed significantly higher 23Na concentrations than IDH wild-type gliomas (median [interquartile range]: IDH wild-type = 52.37 mM [45.98 - 58.56 mM], IDH mutated = 65.02 mM [58.87-67.05 mM], p = 0.039). Among the glioblastomas, there was a trend towards increased 23Na concentration in MGMT methylated tumors that did not reach statistical significance (median [interquartile range]: MGMT methylated = 57.59 mM [50.70 - 59.17 mM], MGMT not methylated = 48.78 mM [45.88 - 53.91 mM], p = 1.0). CONCLUSIONS: 23Na MRI correlates with the IDH mutation status and could therefore enhance image guidance towards biopsy sites as wells as image-guided surgery and radiotherapy. Furthermore, the successive decrease of 23Na concentration from central necrosis to normal-appearing white matter suggests a correlation with tumor infiltration.

Sodium relaxometry using 23 Na MR fingerprinting: A proof of concept.

PURPOSE: To evaluate the feasibility of 23 Na MR fingerprinting (MRF) for simultaneous quantification of T1 , T2l∗ , T2s∗ , T2∗ in addition to ΔB0 . METHODS: A framework for sodium relaxometry using MRF at 7T was developed, allowing simultaneous measurement of relaxation times and inhomogeneities in the static field. The technique distinguishes between bi- and monoexponential transverse relaxation and was validated in simulations with respect to the ground truth. In phantom measurements, a resolution of 2 × 2 × 12 mm3 was achieved within 1 h acquisition time, and the resulting parameter maps were compared to results from reference methods. Relaxation times in five healthy volunteers were measured with a resolution of 4 × 4 × 12 mm3 . RESULTS: Phantom experiments revealed an agreement between the relaxation times obtained via 23 Na-MRF and the reference methods. In white matter, a longitudinal relaxation constant of T1 = 38.9 ± 4.8 ms was found, while values of T2l∗ = 29.2 ± 4.9 ms and T2s∗ = 4.7 ± 1.2 ms were found for the long and short component of the transverse relaxation. In cerebrospinal fluid, T1 was 67.7 ± 6.3 ms and T2∗ = 41.5 ± 3.4 ms. CONCLUSION: This work demonstrates the feasibility of 23 Na-MRF for relaxometry in sodium MRI in both phantom and in vivo studies. Simultaneous quantification of T1 , T2l∗ , T2s∗ , T2∗ and ΔB0 was possible within a 1 h measurement time.

Correction to: Accelerated quantification of tissue sodium concentration in skeletal muscle tissue: quantitative capability of dictionary learning compressed sensing.

The original version of this article unfortunately contained a mistake. Title was incorrect.

Accelerated quantification of tissue sodium concentration in skeletal muscle tissue: quantitative capability of dictionary learning compressed sensing.

OBJECTIVE: To accelerate tissue sodium concentration (TSC) quantification of skeletal muscle using 23Na MRI and 3D dictionary-learning compressed sensing (3D-DLCS). MATERIALS AND METHODS: Simulations and in vivo 23Na MRI examinations of calf muscle were performed with a nominal spatial resolution of [Formula: see text]. Fully sampled and three undersampled 23Na MRI data sets (undersampling factors (USF) = 3, 4.4, 6.7) were evaluated. Ten healthy subjects were examined on a 3 Tesla MRI system. Results of the simulation study and the in vivo measurements were compared to the ground truth (GT) and the fully sampled fast Fourier transform (NUFFT) reconstruction, respectively. RESULTS: Reconstruction results of simulated data with optimized 3D-DLCS yielded a lower deviation (< 4%) from the GT than results of the NUFFT reconstruction (> 5%) and a lower standard deviation (SD). For in vivo measurements, a TSC of [Formula: see text] was observed. The mean deviation from the reference is lower for the undersampled 3D-DLCS reconstructions (3.4%) than for NUFFT reconstructions (4.6%). SD is reduced using 3D-DLCS. Compared to a fully sampled NUFFT reconstruction, acquisition time could be reduced by a factor of 4.4 while maintaining similar quantitative accuracy. DISCUSSION: The optimized 3D-DLCS reconstruction enables accelerated TSC measurements with high quantification accuracy.

Corrections of myocardial tissue sodium concentration measurements in human cardiac 23 Na MRI at 7 Tesla.

PURPOSE: To quantify the tissue sodium concentration (TSC) in cardiac 23 Na MRI. To evaluate the influence of different correction methods on the measured myocardial TSC. METHODS: 23 Na MRI of four healthy subjects was conducted at a whole-body 7T MRI system using an oval-shaped 23 Na birdcage coil. Data acquisition was performed with a density-adapted 3D radial pulse sequence using a golden angle projection scheme. 1 H MRI data were acquired at a 3T MRI system to generate a myocardial mask. Retrospective cardiac and respiratory gating were used to reconstruct 23 Na MRI data in the diastolic phase and exhaled state. B0 and B1 inhomogeneity and partial volume (PV) effects were corrected. Relaxation times and TSC of ex vivo blood samples and calf muscle were determined. These values were used in the PV correction to estimate myocardial TSC, which was compared with the measured TSC of calf muscle. RESULTS: Without any correction the measured myocardial TSC was (54 ± 5) mM. The applied correction methods reduced these values by (48 ± 5)% to (29 ± 3) mM, where PV correction had the largest effect (reduction of (34 ± 1)%). Respiratory and cardiac motion gating decreased the concentrations by (11 ± 1)%. With the applied setup, the corrections of B0 and B1 inhomogeneity (reduction of (3 ± 2)%) had negligible influences on TSC values. The resulting myocardial TSC was approximately 1.4-fold higher than the measured TSC of calf muscle tissue of the same healthy subjects ((20 ± 3) mM). CONCLUSION: For quantitative human cardiac 23 Na MRI several corrections are needed and ranked for our setup: PV correction, respiratory and cardiac gating, correction for B1 inhomogeneity effects.

High-resolution FLAIR MRI at 7 Tesla for treatment planning in glioblastoma patients.

Ultra-high field MRI is an emerging technique promising high-resolution images for radiotherapy planning. We compared a 7 Tesla FLAIR sequence with clinical FLAIR imaging at 3 Tesla in glioblastoma patients before radiotherapy. High-resolution 7 Tesla FLAIR imaging may enhance the depiction of organs at risk and possibly modify target volumes.

Probing the microscopic environment of 23 Na ions in brain tissue by MRI: On the accuracy of different sampling schemes for the determination of rapid, biexponential T2* decay at low signal-to-noise ratio.

PURPOSE: To investigate and to reduce influences on the determination of the short and long apparent transverse relaxation times ( T2,s*, T2,l*) of 23 Na in vivo with respect to signal sampling. METHODS: The accuracy of T2* determination was analyzed in simulations for five different sampling schemes. The influence of noise in the parameter fit was investigated for three different models. A dedicated sampling scheme was developed for brain parenchyma by numerically optimizing the parameter estimation. This scheme was compared in vivo to linear sampling at 7T. RESULTS: For the considered sampling schemes, T2,s* / T2,l* exhibit an average bias of 3% / 4% with a variation of 25% / 15% based on simulations with previously published T2* values. The accuracy could be improved with the optimized sampling scheme by strongly averaging the earliest sample. A fitting model with constant noise floor can increase accuracy while additional fitting of a noise term is only beneficial in case of sampling until late echo time > 80 ms. T2* values in white matter were determined to be T2,s* = 5.1 ± 0.8 / 4.2 ± 0.4 ms and T2,l* = 35.7 ± 2.4 / 34.4 ± 1.5 ms using linear/optimized sampling. CONCLUSION: Voxel-wise T2* determination of 23 Na is feasible in vivo. However, sampling and fitting methods have to be chosen carefully to retrieve accurate results. Magn Reson Med 80:571-584, 2018. © 2018 International Society for Magnetic Resonance in Medicine.

Reproducibility of CMRO2 determination using dynamic 17 O MRI.

PURPOSE: To assess the reproducibility of 17 O MRI-based determination of the cerebral metabolic rate of oxygen consumption (CMRO2 ) in healthy volunteers. To assess the influence of image acquisition and reconstruction parameters on dynamic quantification of functional parameters such as CMRO2 . METHODS: Dynamic 17 O MRI data were simulated and used to investigate influences of temporal resolution (Δt) and partial volume correction (PVC) on the determination of CMRO2 . Three healthy volunteers were examined in two separate examinations. In vivo 17 O MRI measurements were conducted with a nominal spatial resolution of (7.5 mm)3 using a density-adapted radial sequence with golden angle acquisition scheme. In each measurement, 4.0 ± 0.1 L of 70%-enriched 17 O gas were administered using a rebreathing system. Data were corrected with a PVC algorithm, and CMRO2 was determined in gray matter (GM) and white matter (WM) compartments using a three-phase metabolic model (baseline, 17 O inhalation, decay phase). RESULTS: Comparison with the ground truth of simulations revealed improved CMRO2 determination after application of PVC and with Δt ≤ 2:00 min. Evaluation of in vivo data yields to CMRO2,GM = 2.31 ± 0.1 µmol/g/min and to CMRO2,WM = 0.69 ± 0.04 µmol/g/min with coefficients of variation (CoV) of 0.3-5.5% and 4.3-5.0% for intra-volunteer and inter-volunteer data, respectively. CONCLUSION: This in vivo 17 O inhalation study demonstrated that the proposed experimental setup enables reproducible determination of CMRO2 in healthy volunteers. Magn Reson Med 79:2923-2934, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Iterative reconstruction of radially-sampled 31P bSSFP data using prior information from 1H MRI.

The purpose of this study is to improve direct phosphorus (31P) MR imaging. Therefore, 3D density-adapted radially-sampled balanced steady-state free precession (bSSFP) sequences were developed and an iterative approach exploiting additional anatomical information from hydrogen (1H) data was evaluated. Three healthy volunteers were examined at B0=7T in order to obtain the spatial distribution of the phosphocreatine (PCr) intensities in the human calf muscle with a nominal isotropic resolution of 10mm in an acquisition time of 10min. Three different bSSFP gradient schemes were investigated. The highest signal-to-noise ratio (SNR) was obtained for a scheme with two point-reflected density-adapted gradients. Furthermore, the conventional reconstruction based on a gridding algorithm was compared to an iterative method using an 1H MRI constraint in terms of a segmented binary mask, which comprises prior knowledge. The parameters of the iterative approach were optimized and evaluated by simulations featuring 31P MRI parameters. Thereby, partial volume effects as well as Gibbs ringing artifacts could be reduced. In conclusion, the iterative reconstruction of 31P bSSFP data using an 1H MRI constraint is appropriate for investigating regions where sharp tissue boundaries occur and leads to images that represent the real PCr distributions better than conventionally reconstructed images.

Three-dimensional dictionary-learning reconstruction of (23)Na MRI data.

PURPOSE: To reduce noise and artifacts in (23)Na MRI with a Compressed Sensing reconstruction and a learned dictionary as sparsifying transform. METHODS: A three-dimensional dictionary-learning compressed sensing reconstruction algorithm (3D-DLCS) for the reconstruction of undersampled 3D radial (23)Na data is presented. The dictionary used as the sparsifying transform is learned with a K-singular-value-decomposition (K-SVD) algorithm. The reconstruction parameters are optimized on simulated data, and the quality of the reconstructions is assessed with peak signal-to-noise ratio (PSNR) and structural similarity (SSIM). The performance of the algorithm is evaluated in phantom and in vivo (23)Na MRI data of seven volunteers and compared with nonuniform fast Fourier transform (NUFFT) and other Compressed Sensing reconstructions. RESULTS: The reconstructions of simulated data have maximal PSNR and SSIM for an undersampling factor (USF) of 10 with numbers of averages equal to the USF. For 10-fold undersampling, the PSNR is increased by 5.1 dB compared with the NUFFT reconstruction, and the SSIM by 24%. These results are confirmed by phantom and in vivo (23)Na measurements in the volunteers that show markedly reduced noise and undersampling artifacts in the case of 3D-DLCS reconstructions. CONCLUSION: The 3D-DLCS algorithm enables precise reconstruction of undersampled (23)Na MRI data with markedly reduced noise and artifact levels compared with NUFFT reconstruction. Small structures are well preserved.

Amide proton transfer of carnosine in aqueous solution studied in vitro by WEX and CEST experiments.

Amide protons of peptide bonds induce an important chemical exchange saturation transfer (CEST) contrast in vivo. As a simple in vitro model for a peptide amide proton CEST effect, we suggest herein the dipeptide carnosine. We show that the metabolite carnosine creates a CEST effect and we study the properties of the exchange of the amide proton (-NH) of the carnosine peptide bond (NHCPB) in model solutions for a pH range from 6 to 8.3 and a temperature range from T = 5 °C to 43 °C by means of CEST and water exchange spectroscopy (WEX) experiments on a 3 T whole-body MR tomograph. The dependence of the NHCPB chemical exchange rate k(sw) on pH and temperature T was determined using WEX. For physiological conditions (T = 37 °C, pH = 7.10) we obtained k(sw) = (47.07 ± 7.90)/s. With similar chemical shift and exchange properties to amide protons in vivo, carnosine forms a simple model system for optimization of CEST pulse sequences in vitro. The potential for direct detection of the metabolite carnosine in vivo is discussed.

Iterative 3D projection reconstruction of (23) Na data with an (1) H MRI constraint.

PURPOSE: To increase the signal-to-noise ratio (SNR) and to reduce artifacts in non-proton magnetic resonance imaging (MRI) by incorporation of a priori information from (1) H MR data in an iterative reconstruction. METHODS: An iterative reconstruction algorithm for 3D projection reconstruction (3DPR) is presented that combines prior anatomical knowledge and image sparsity under a total variation (TV) constraint. A binary mask (BM) is used as an anatomical constraint to penalize non-zero signal intensities outside the object. The BM&TV method is evaluated in simulations and in MR measurements in volunteers. RESULTS: In simulated BM&TV brain data, the artifact level was reduced by 20% while structures were well preserved compared to gridding. SNR maps showed a spatially dependent SNR gain over gridding reconstruction, which was up to 100% for simulated data. Undersampled 3DPR (23) Na MRI of the human brain revealed an SNR increase of 29 ± 7%. Small anatomical structures were reproduced with a mean contrast loss of 14%, whereas in TV-regularized iterative reconstructions a loss of 66% was found. CONCLUSION: The BM&TV algorithm allows reconstructing images with increased SNR and reduced artifact level compared to gridding and performs superior to an iterative reconstruction using an unspecific TV constraint only.