Figure 5. Estimated metabolite ADC maps from the 64x64x12 data at a nominal resolution of 3.4×3.4×5.3 mm³. Row 1 shows the T1-weighted images for several slices from the 3D imaging volume. Rows 2-4 display the ADC maps of NAA, Cr, and Cho (from Gdir1) for the corresponding slices, respectively. Patterns of white matter and gray matter diffusion property differences (white matter having larger metabolite ADCs than gray matter) can be visualized due to the high resolution. To the best of our knowledge, these are the highest resolution metabolite ADC maps ever produced.

Figure 4. Reconstructed DW metabolite and MD maps from the 32x32x8 data. Columns 1-3 present the metabolite maps at 3 b-values from Gdir1. Column 4 shows the MD maps of the matched slice for NAA, Cr, and Cho respectively (Rows 1-3). The MD maps are generated by averaging the registered ADC maps fitted from 3 orthogonal diffusion encoding directions and then overlaid on aligned anatomical images. DW contrast can be observed and the ranges and values of the MD for each metabolite are consistent with reported values^7-9,17.

Figure 2: Exemplary concentration estimate maps of all successfully quantified metabolites in one subject. For lower SNR-metabolites, more artifacts or regions without successful quantification are visible. Good visibility of GM/WM and regional variations can be seen especially for neurotransmitters.

Figure 3: Mean concentration estimates per ROI in [mM/l] for successfully quantified metabolites in all ROIs qualified as defined in Fig.1.

Figure 1: Top: Schematic diagram of the fMRS visual stimulation paradigm using a red-black checkerboard pattern. Numbers indicate the acquired number of averages per block and needed acquisition time. FLASH insets below show the BOLD activated region and the fMRS voxel placement (15x18x20 mm³).

Bottom: Representative MR spectra from one volunteer acquired during STIM (red, S1.2/S2.2) and REST (blue, R2.2/R3.2) periods (64 averages each spectrum) and the difference spectrum (yellow). Small amplitude and linewidth changes of Cr+PCr and NAA indicate the BOLD effect.

Figure 4: Left: Mean time courses of Lac and Glu concentration differences relative to a baseline concentration (R1.2) with a time resolution of 40 s. Error bars represent standard deviations of the mean over all volunteers.

Right: Correlation plots between the metabolite concentrations (from spectra summed across all volunteers) and the MC water amplitude (blue crosses). Solid red lines represent the linear fit, dotted red lines 95% confidence intervals. R and p values indicate the Spearman’s Rank Correlation Coefficient and significance level of the correlation, respectively.

FIGURE 2 Time-course spectra and corresponding fitted spectra of Glu, Gln, and their ¹³C satellites acquired from the dACC of subject 1. No linebroadening was applied to the spectra. Voxel size = 3.5 × 1.8 × 2 cm³; TR = 2.2 s; TE = 56 ms; number of averages = 264 and total scan time = 10 min for the pre-¹³C spectra; number of averages = 132 and total scan time = 5 min for each individual post-¹³C spectrum.

FIGURE 4 Plots of fractional enrichments of Glu H4 and Gln H4 vs. time after oral administration of [U-¹³C]glucose for all five healthy volunteers.

Figure 4. Results obtained from a fixed rat brain injected with PCE and PFOB NEs. PCE (red) and PFOB (green) maps are overlaid on the ¹H anatomical image, and representative spectra are shown. As can be seen, the proposed method obtained PCE and PFOB maps and spectra as expected.

Figure 3. Proton structural image, field map, and comparison of Fourier-based and proposed ¹⁹F-MRSI processing results obtained from a phantom. As can be seen, Fourier-based processing results suffered from spatial chemical shift artifacts and spectral aliasing artifacts, while the proposed method produced artifact-free spatiospectral functions for PCE and PFOB.

A. Representative neurometabolic profiles in BDL and Sham rats acquired at 14.1T in the cerebellum (STEAM sequence - TE=3ms, TM=10ms, LB=8Hz) B. Representative spectra at different b-values for BDL and SHAM rats acquired with the STE-LASER⁴ sequence (LB=8Hz) in the cerebellum at 14.1T. C. Individual metabolite decays versus b-values for both groups. Error bars: dispersion around the mean value. Statistics: 2-way ANOVA with variables disease and b-value, *: p<0.05.

IHC staining in the cerebellum of Sham and BDL rats. Top: anti-GFAP(red)/DAPI(blue) staining of the cerebellar folium, scale bar: 100μm. Middle: GFAP/DAPI staining of the granular layer, scale bar: 25μm. Bottom: Colgi-Cox staining of Purkinje cells, scale bar: 25μm Right: corresponding quantifications, ** p<0.01, *** p<0.001, **** p<0.0001.

Fig. 1: Metabolic model: Glucose is transported through the blood-brain barrier by GLUT-1 transporters and metabolized to pyruvate/lactate by CMR_gl. Lactate also enters via monocarboxylic acid transporters, characterized by V_max, K_M and K_D. Lactate/pyruvate is further metabolized in the TCA cycle (V_tca). Unlabeled substrate in- and outflows are V_dil,gly = V_{lac,out,brain} and V_dil,tca. The glutamate/glutamine cycle is represented by V_gln. Modifications for ²H are indicated in red. ¹³C and ²H label positions are shown.

Fig. 3: Axial MR image for spatial assignment. Time course of ²H MR spectra during [6,6’-²H₂]-glucose infusion, starting from baseline prior to infusion until the end of the experiment with n=64 averages each, acquired from combined cortex/subcortex (Cx/SCx) volume. Line broadening=2Hz. Labeled metabolites are marked as double labeled but the signals include both single and double ²H-labeled metabolites.

Figure 1. Schematic of qMRS experiment. Subject drinks aqueous deuterated glucose solution. Prior to this, subject has undergone MR spectroscopy and chemical shift imaging (qCSI) at 7T. After glucose ingestion, subject is scanned again. In this post-glucose scan, 1H resonances of metabolites experiencing labeling downstream from glucose will decrease. qCSI measurements are made repeatedly and, as labelled glucose is absorbed from the bloodstream and metabolized, the fractional enrichment of detected metabolites increases and eventually plateaus.

Figure 3. Timecourse qCSI data of [Glu] and [NAA] from all subjects, gray and white matter. A) Glu/NAA ratio map (MRSI) illustrating ROIs used for calculation of the plotted data designated as gray and white matter regions. B) Bar graphs showing the timecourse measurements of Glu and NAA by qCSI in gray and white matter. In this plot, individual dots correspond to a single voxel (out of 15) in each ROI, averaged across four subjects who underwent this part of the experiment.

Figure 1. Osprey analysis workflow to remove residual metabolite signals and generate ‘clean’ MM spectrum.

Figure 2. MM spectra from centrum semiovale white matter. Spectra from 96 subjects are overlaid in panel A, and the metabolite model signals are labeled in panel B. The mean spectra for male and female subjects are compared in panel C. The mean spectra for each decade of age are compared in panel D.

Figure 1. Example of multivoxel MRS and ROIs from a patient with pre-manifest HD.

Figure 3. Significant differences in metabolite ratios between either the PM or EM groups and all of controls combined (HC_EM+HC_PM). Metabolites from healthy controls were plotted for each subgroup (HC_EM, HC_PM). When compared to age-/gender-matched healthy controls, only glutathione/creatine within the insula (EM vs. HC_EM, p=0.002) remained significant.

Figure 2: a. Spectra without diffusion-weighting obtained with the STE-LASER sequence (blue dotted line), the broad pulse SE-LASER sequence (green line) and the polychromatic SE-LASER sequence (red line). b. The gain of lactate signal is spectacular when using the polychromatic SE-LASER sequence.

Figure 3: STE-LASER and polychromatic SE-LASER data at high diffusion-weighting in WT mice. a. Spectra acquired up to high diffusion-weighting (from b = 0.02 ms/µm² to 20 ms/µm²) with the STE-LASER sequence (left) and the polychromatic SE-LASER sequence (right). b. LCModel analysis for metabolites of interest. Macromolecules contribution on 1.31 ppm lactate peak is noticeable. c. Signal attenuation as a function of b for three metabolites by using the two sequences (blue diamonds = polychromatic SE-LASER, red squares = STE-LASER). Data points and error bars stand for mean ± s.d.

Figure 2: In vivo MEGA-PRESS experiments at TE 90 and 110 ms for editing the PE_3.22 (left) and PE_3.98 (right). NAA: N-acetylaspartate; Lac: lactate; Cho: choline; and Cr: creatine.

Figure 3: Linear-combination modeling of PE_3.22 and PE_3.98 difference spectra with complete fit (yellow), spline baseline (black), and PE estimates at both TEs 90 ms (blue) and 110 ms (red). Asp: aspartate, Cho: choline, Cr: creatine, Gln: glutamine, Ins: myo-Inositol, Lac: lactate, MM: macromolecules, Tau: taurine.

FIg. 2. (A,B) Spectral data from the frontal-parietal (A) and temporal (B) regions. CRLB % are shown for several loci in the temporal study. (C1,C2) The GLU/tNAA regresses significantly with fraction GM from both studied regions. (D) The CRLB % are shown for GLU, INS and GLN for the frontal-parietal and temporal regions.

Fig. 3. (A) Regression of CRLB % with ξ_k=√LW/SNR_k from the frontal-parietal region. (B) Same as (A) but with superimposed data from the temporal region, and magnified in X and Y axes. (C) Plots of CR_k = CRLB_k% * Amp_k, with ξ₀ =√LW*σ, with different compounds. The regressions for CRE, INS and CHO are shown.

Representative spectrum in the ACC and PCC. Volumes of interest of magnetic resonance spectrum were placed at the ACC and PCC. A representative MRS acquired with the SPECIAL sequence, the corresponding LCModel spectral fit, fit residual, macromolecules (MM), baseline and individual metabolite fits including GSH in the ACC and PCC.

a) Sagittal, coronal, and transverse brain views showed voxels with positive associations between apathy scale scores and ¹⁸F-PM-PBB3 SUVRs in patients with PSP after controlling for the effects of age and PSPRS. b, c) Scatterplots showing the association between apathy scale scores and ¹⁸F-PM-PBB3 SUVRs in the detected areas (AG).

Figure 2. a) Subject 1: mean lactate concentration across sleep-wake cycles decline during sleep relative to wakefulness: 4.6 % in N1 (2.5 min), 12.0 % in N2 (12 min), and 18.0 % in N3. b) Representative sleep stages were visually scored for 30 s epochs and classified into: Wake (W), Non-REM sleep stage 1 (N1), Non-REM sleep stage 2 (N2), and Non-REM sleep stage 3 (N3).

Figure 1. a) Measured single voxel (22 cm3, orange square) on a sagittal anatomical MR image of a 27-year old healthy young female (subject 1). b) A 7.5 sec 3T single voxel 1H-MR spectroscopy showing metabolites including lactate, c) Lactate concentrations, calculated within [1.23-1.48 ppm], are averaged over each sleep stage including Wake (W), non-REM stage 2 (N1), non-REM stage 2 (N2) and non-REM stage 3 (N3).

Fig. 2. Activations in the a priori defined ROI of the bilateral gyrus fusisormis for face-stimuli versus non-face stimuli (a, N = 18, MNI peak voxel = 38, -38, -18, p_uncorr < 0.001, p_fwe < 0.001) and correlations of these activations with hallucination proneness in the face task (b, negative correlation, N = 18, MNI peak voxel = 38, -38, -18, p_uncorr < 0.001, p_fwe = 0.040) and glutamate/NAA ratios obtained from MRS (c, N = 16, MNI peak voxel = 32, -40, -16, p_uncorr < 0.001, p_fwe = 0.121). In the figure, p-values of 0.005 were used for illustration purposes.

Fig. 1. ¹H difference MR spectra from a VOI in the right visual cortex (bottom inset) acquired at 3T with the MEGA-PRESS sequence (TR/TE = 3000/68 ms) for a healthy volunteer together with LCModel fit, LCModel output for GABA, Glu, fit residuals, and background. Preprocessing of data included coil combination, removal of bad averages, and phase and frequency correction.

Fig.3: Fitted diffusion decays as obtained with the different approaches described in the text. Free-Fit data was modeled as mono- and biexponential decay. Biexp-Fit (blue) is consistent for some metabolites but shows diverging results for others. The non-Gaussian nature of diffusion is evident in both approaches for most metabolites. At the bottom, spectra with the fits from Free-Fit as well as Biexp-Fit are shown documenting how small the differences are even though the decay function seem to differ substantially for some metabolites. (blue: Biexp-Fit; red; Free-Fit)

Fig. 1: Illustration to motivate the need for the MM peak-based signal correction. For multiple subjects, the peak areas of MM0.9 are plotted in relation to the monoexponential decay fit for the cohort data from subjects without apparent strong artifactual signal decay. Subject 2 shows systematic additional signal decay with higher b-values on top of the MM diffusion effect, while subjects 1 and 3 show strong random fluctuations.

Spatially-localized spectra for a 4x4 voxel array from an anterior brain region are compared, with results produced by the standard reconstruction (see texts) and the proposed method (the blue curves). The location of the region-of-interest is indicated by the red box in the anatomical image on the left with overlaid voxel grids. Significant reduction in lipid leakage and SNR improvement can be observed for the spectra from the proposed method, especially for the edge voxels. Spectra are shown in real parts.

Comparison of NAA/Cr (column 1), Cho/Cr (column 2), Glu/NAA (column 3) maps from the standard reconstruction and the proposed method for one subject. As can be seen, the metabolite ratio maps produced by the proposed method have less artifacts (examples indicated by the red arrows) and better gray/white matter contrast (e.g., the Glu/NAA maps). The color scales were the same for the same column. The metabolite levels were obtained using spectral integration for this preliminary study, and no relaxation correction was performed.

Figure 3: Human scan results. a) Reference image with the scan location, b) water image with the same parameters as the spectroscopic imaging scan. c) ¹H spectra at three regions shown in b). The scan parameters are described in the main text.

Figure 4: 3D-shaped phantom imaging a) RS-EPSI versus RS-COKE with 3 readout segments with the “brain only” phantom. Images show water and NAA maps and the spectrum for the central box (5x5 pixels). b) 3D-shaped phantom including brain-like compounds and a superficial compartment with a lipid-like compound. The slice location is shown by blue overlay. Images compare single and 3-segment water images, a 3-segment acquisition with water suppression, and a 3-segment acquisition with both water suppression and three saturation bands. Sequence parameters are given in the main text.

Figure 1: FIDESI features two analog-to-digital-converter readouts: Large k-space coverages for the high SNR FID scan and a small k-space coverage for the low SNR echo scan at TE 288 ms. The crusher gradient superimposed on the localization gradient ramp-down caused spectral acoustic sideband artifacts. A special pair of gradients was played out between the GOIA pulses for nulling the zero-order magnetization caused by the initial CRT gradients which move the trajectory to the specific ring radii.

Figure 4: Whole brain metabolic Maps of the healthy Volunteer. FID maps have an effective measured matrix size of 64x64x29, echo maps have an effective measured matrix size of 32x32x15, but are interpolated to 32x32x29 before quantification. All maps are additionally interpolated by a factor of 2 for visualization purposes.

Figure 2: Time series of the MC-semiLASER spectra from the occipital lobe (left) and frontal cortex (right) and the corresponding difference spectra underneath. The difference spectra in the spectral region of interest between 1.9 and 2.8 is zoomed in to make changes more visible. The colorbar indicates the time of the (difference) spectra after the glucose administration. Changes in the spectral region of NAA and Cr can occur due to subtraction errors.

Figure 3: a) ¹H MC-semiLaser spectrum from one volunteer of the occipital lobe before glucose administration (black line) and 127 minutes after the administration (red line) and the difference spectrum (blue line). Obvious peak changes are marked: Decreasing peaks are marked in purple and increasing peaks are marked in green. The spectra contain 64 averages. b) Simulated spectra for selected metabolites and labeled metabolites.

Figure.1 Transformation of the Hadamard reconstructed spectra (SUM - green, diff_GABA - red, diff_GSH - yellow) into a single concatenated spectrum with 1.2 ppm gap separation

Figure 3. Mean processed spectral data is shown for test (black) and retest (grey) acquisitions at the top; mean LCModel fits, residuals and baselines are presented for test (black) and retest (grey) at the bottom

Figure 3: Metabolite spectra obtained from the occipital lobe (top row) and parietal lobe (bottom row) for the three water suppression schemes tested here, VAPOR, COWS(7;236) and COWS(12;626). COWS(7;236) has comparable performance to VAPOR, while COWS(12;626) has marked improvement over both. Note that WS schemes were applied as theoretically optimized and no experimental fine tuning was performed.

Figure 4: Macromolecule spectra obtained from the parietal lobe for VAPOR (red) and COWS(7;236) (black). Marked reduction of residual water for COWS(7;236) was observed over VAPOR. COWS(12;626) was not used due to its interference with the double inversion preparation module.

Figure 1. Schematic layout of the experimental design for ¹H-[¹³C] MRS acquisition of both hippocampus and hypothalamus with dynamic shim update (DSU) (a), and typical non-edited (b) and edited (c) MR spectra of hippocampus (blue) and hypothalamus (orange).

Figure 3. Neurochemical profiles of hippocampus and hypothalamus

Figure 2. Baseline 37°C spectrum after PWS to 1% residual water. Good fidelity observed for nearest neighbors, the C1α of glucose and the anomeric proton of NAA. Spectral baseline for normal fit region above 4.2 ppm is unperturbed.

Figure 4. 37C Circulatory arrest. Powder-pattern develops during circulatory arrest and resolves after 5 minutes of recovery. The residual water reveals low level powder pattern which provides a measure of deoxy-Hb and hypoxia. Glucose recovers but lactate persists after hypoxia reversed.

Pipeline of volumetric brain tissue segmentation from MRI and MRSI data. A. A binary mask of a MRSI single slice VOI (8x10 cm2) was created using the SPM toolbox. B. partial volume masks for each tissue type were created using FSL FAST. C. Partial volume segmentation of the lesion filled T1-MPRAGE structural images. D. Tissue segmentation of MRSI VOI (CSF, GM and WM) and lesion segmentation overlaid on the T2-FLAIR image.

A. Box plot of significant differences in the neurometabolic ratios (NAA/tCr)and (m-Ins/tCr) for WML voxels, NAWM-MS and HC voxels in four slices of one DMR, located within deep cortical white matter in posterior parietal lobes (left). B. Mapping of NAA/tCr ratios in single slice and multi-slice of 3 DMRs located in one axial brain slice of an MS patient, using spiral-MRSI based representation of data, overlayed with structural image.

Figure 3: Temporal Glx increases are observed in saline-injected STA animals, showing both between group and baseline significance. Baseline significance begins 75 min post-injection, slightly before behavioral migraine symptoms present in the nitroglycerin (NTG) group. NTG (10 mg/mg) was injected at t=0 min. Brackets indicate significance between groups (p<0.05); * is significance to baseline (p<0.05).

Figure 2: Periorbital Von Frey mechanical sensitivity thresholds (force based on filament stiffness) for both (a) Naturally Resilient (NR) and (b) Spontaneous Trigeminal Allodynic (STA) rats

Figure 2 Visualization of the significant differences in relative metaboliteslevels between patients with SCA1 and controls in the three VOIs

Figure 3 Correlation between relative metabolite levels in the brain and clinical SARA scores. tNAA is significantly correlated in all three VOIs, glutamate is correlated with clinical scores in the cerebellar white matter.

Figure 2. Raincloud plot depicting the GSH concentration distribution (mmol/l) in the anterior cingulate cortex of PPCS participants with intermittent (mean = 2.31 ± 0.41; n = 6) and constant headache (mean = 2.86 ± 0.61; n = 10) and matched, injury-free controls (mean = 2.36 ± 0.37; n = 13).

Figure 3. Association between functional impact of headache and GSH concentration (mmol/l) in the anterior cingulate cortex. GSH in the anterior cingulate cortex was positively correlated with HIT-6 scores; (higher HIT-6 scores indicates greater functional impact of headache) r(16) = 0.507, p = 0.045*. HIT-6 score classifications: (1) little-to-no impact (score: 36-39), (2) moderate impact (score: 50-55), (3) substantial impact (score: 56-59), (4) severe impact (score 60-78).

Figure 1: Representative spectroscopic map images for a single control subject. Presented is every fifth slice from the volumetric data for (A) creatine (Cr), (B) choline (Cho), (C) myo-inositol (mI), (D) glutamate plus glutamine (Glx), (E) N-acetyl aspartate (NAA), (F) grey and (G) white matter segmentation (GM, WM), (H) lobar parcellation, and (I) water (used as quantification reference) at the spectroscopic image spatial resolution.

Figure 3: Boxplots of lobar metabolite distributions within mTBI and control (CTL) cohorts. Occipital WM Cho and Cr, parietal WM Cr, and frontal WM mI are higher in mTBI patients compared to controls (MW, ♦: p 0.05). Note that in all lobes WM Cr and Cho medians are higher in mTBI patients compared to controls. Cho: choline, Cr: creatine, mI: myo-inositol, Glx: glutamine plus glutamate; NAA: N-acetyl aspartate; WM: white matter.

Figure 1. On left: Representative 2.5 x 2.5 x 2.5 cm3 voxel in occipital-parietal sulcus overlaid onto 7T MPRAGE anatomical image indicating region of MRS acquisition. Image and voxel reconstruction was performed prior to the follow-up scan, to ensure consistent placement of MRS voxel within participants. On right: fits from participant S01 for the two 7T and six 3T spectra spectra acquired in the voxel of interest.

Figure 3. Pearson’s correlation coefficient (r) computed for each metabolite between 3T sequences shown in the x-axis and 7T sequences in legend. These were then averaged across all metabolites within a 3T-7T pairing to obtain the sequence-wide average correlation between 3T measures and 7T measures of metabolites.

Figure 1. Overview of the synthetically altered spectra and their LCModel outputs for one representative subject. From top to bottom: synthetically altered spectra, fitted baselines, fitted GABA signals and residuals. The overlap of GABA resonances with other metabolite is shown with different colors. The alterations of the spectra caused by changes in GABA concentration are visible at 1.90 and 2.29 ppm whereas almost invisible at 3.00 ppm. The LCModel residuals do not show any structured signals for any of the spectra

Figure 3. [GABA] values corresponding to the synthetically altered spectra (A) for different LCModel approaches and (B) for different number of averages are plotted as a function of the imposed change in GABA signal. The unity line between imposed change and measured concentration is also plotted. The results showed that the approach #1 (measured MM, stiff baseline and no concentration ratio priors) allows to correctly quantify the imposed change in GABA concentration. Reducing number of averages does not influence the mean value of [GABA] in a cohort but increase SD.

Figure 2: Representative B0 maps for both shim settings (once only with scanner shim of up to second order spherical harmonics and once with the 24-channel local shim array) are shown in three orientations overlaid on anatomical images for a single volunteer. B0 maps are plotted between [-100Hz,100Hz]. The standard deviation of the frequency shifts are reported under each corresponding B0 map.

Figure 4: Representative map of glutamate and glutamine (Glu+Gln) for the standard scanner shim and the 24-channel shim, along with the corresponding B0 maps. Regions that suffer from poor B0 shimming quality with the scanner shim are marked by red arrows. The B0 maps are plotted between -60 Hz (blue) and 60 Hz (red) and the corresponding standard deviations are noted below each B0 map.

FIGURE 1. MR spectra acquired from all subjects and voxel position (3.0×3.0×3.0 (cm)³ on 3D MPRAGE T1-weighted images (TR/TI = 2300/1050 ms)). The spectra baseline-corrected on LCModel are stacked with SNR of the H1-α-glucose peak at 5.23 ppm in semi-LASER spectra from low (bottom) to high (top). The downfield spectra (5.0-5.4 ppm in the right panel) are scaled with 5-fold. In blue short-TE STEAM spectra, the glucose peak was not identified. In red spectra acquired from a subject, the glucose peak could not be fitted due to strong nuisance signals between the peaks of glucose and water.

TABLE 1. Results from the spectral analysis.

Figure 1. 3T MRI scanner setup for neonatal pig DHCA-CPB study. A coronal T₁-weighted MRI with the selected MRS voxel and representative spectra at baseline and end circulatory arrest (CA) time points are shown.

Figure 4. Observation of NMR “powder pattern” from residual water signal from a brain voxel in a neonatal pig model under circulatory arrest. Dynamic spectra are plotted as a function of time post-circulatory arrest. A powder pattern is theoretically predicted from the intravascular water signal, based on a model of randomly oriented veins/capillaries containing deoxyhemoglobin [5].

Figure 2:Correlations between the three versions of macromolecule-suppressed GABA levels and age. a) Correlation between macromolecule-suppressed GABA and age. b) Correlation between α-corrected macromolecule-suppressed GABA and age. c) Correlation between macromolecule-suppressed GABA/Cr and age. TH: Thalamus (red); SM: Sensorimotor Cortex (green); OC: Occipital Cortex (blue); MM Supp : macromolecule-suppressed.

Table 1: Correlations between macromolecule-suppressed GABA measures and age (months)

Figure 2: Boxplot showing the average concentration of tNAA, tCr, tCho, and Glx in both concussion and OI groups. There are no significant differences between concussion and OI groups in the aforementioned metabolites utilizing multiple ANCOVAs while controlling for age, sex, and site. Circles outside of the box plots are outside the 1.5 standard deviations interquartile range, and the stars are 3 standard deviations outside the interquartile range. All outliers meet all data quality criteria.

Figure 1: Example spectrum of a pediatric participant with concussion. This 1H-MRS PRESS data is from a 2x2x2 cm³ voxel in the left dorsolateral prefrontal cortex, shown in the voxel placements on the anatomical brain scan, to measure tNAA, tCr, tCho, and Glx. The PRESS sequence had a TE of 30ms and a TR of 2000ms, with 96 averages.

Fig.2. Overview of the signals averaged over groups (NORM and mTBI) at 7 ppm and approximation of the signals.

Fig.1. Voxel location: posterior cingulate cortex. Size 50 x 25 x 25 mm.

Fig. 5. Regression plots showing fit to PCSS composite score vs. metabolite concentration. Regression p-value and R2 are indicated.

Fig. 3. Table of significant regressions between neurometabolites and PCSS composite scores. Empty cells indicate that no significant regression was found. The p-value of the overall regression F-statistic is reported along with R2. VO=Vestibular-ocular

Figure 1. Mean voxel location across participants and conditions (top left). Example spectrum (bottom left). Graph representing glutamate ANOVA results. Bars represent the group mean for within condition rest (light blue) to recall (dark blue). Individual connected dots and lines represent within participant rest to recall change (right).

Comparison of Neurochemical profiles (µmol/g) acquired within the hypothalamus of sheep at 4 time points (P01, P02, P03 and P04) during long days (Long period, LP) and at 4 time points during short days (Short period, SP). * p< 0.05; ** p<= 0.01

A. Neurochemical profiles (µmol/g) during Long days (LP) at 4 time points. * p<0.05

B. Neurochemical Profile (µmol/g) during Short days (SP) at 4 time points. * p<0.05 Gln and Glu concentrations at P01 were significantly different compared to concentrations at later time points

Figure 3: MRS acquisition and spectral analysis. (A) Placement of the 3.4 mL voxel (26 x 10 x 13 mm³) in the left hippocampus (52 years, woman), containing mainly its body and tail. (B) Spectrum from the hippocampus fitted with linear combination modelling. Note, no macromolecular background in the residual signal. (C) Basis set used for quantification of hippocampal metabolites.

Figure 1: Simplified simulations of T₂ relaxation of normalized signals for macromolecules (MM) and metabolites. The average T₂ relaxation times and ranges of macromolecules and metabolites were calculated from literature^3,14,15. Note that at T_E = 120 ms, most MM signal is gone, while on average 60% of the metabolite signal remains. This demonstrates that relatively little metabolite signal was lost while the specific choice of T_E = 120 ms enabled easily quantifiable Glu+Gln (Glx)) and m-Ins resonances.

In vivo ³¹P-MR spectra from whole brain (VOI=5x9x9mm³) acquired at week 6 after BDL or sham surgery, each spectrum is from one animal.

In vivo ¹H-MR spectra from hippocampus (VOI=2x2.8x2mm³) acquired at week 6 after BDL or sham surgery, each spectrum is from one animal. Spectra clearly show increased tCr in both BDL+Cr and Sham+Cr. The positive effect of Cr treatment on other osmolytes (Ins, Tau and tCho) is also visible on BDL+Cr spectrum.

Figure 4A. MRSI spectra from a subject with some misalignment between MRI and MRSI. (A) The approximate location of the lipid mask indicated with green voxels. Lipid signal strength is indicated by an orange color map. Without lipid auto-alignment, greater and asymmetric residual lipid signal are clearly visible (4A left). After lipid auto-alignment, the lipid removal is highly symmetric and the lipid residuals are smaller (4A right). The estimated displacement was 4.9 mm and 2.5 mm in X and Y directions, respectively. Spectral range is [0, 4.2] ppm.

Figure 2. Results of spatial auto-alignment in 141 scans. Auto-alignment was performed on multiple scans of 73 healthy subjects. Red circles indicate positions outside a 5mm boundary (defined as Large Displacement). Results showed that coronal head movement was greater than sagittal during scan sessions, consistent with positioning of the head within the RF coil in supine position. The alignment precision exceeded the MRSI nominal voxel size of 20x20x25 mm³. Multi-scale spatial alignment used 7x7 search grids (40.0mm, 13.5mm, 4.5mm, and 1.5mm) to efficiently locate the minimum.

Fig.2: a) Measured spectra between 0.0 and 4.2 ppm for all pulses from session 1_1 for subject 2 b) Bland-Altman plots for HS (blue), GOIA (orange), WURST (green) for R₀ (top), R_1,M (middle), and R_1,W (bottom) to visualize repeatability (R₀) and reproducibility (R_1,M and R_1,W). One point is generated by using the two equations in the main text. The SD indicates the measure of repeatability/reproducibility and is, averaged over all pulses: 0.016, 0.039, and 0.064 for R₀, R_1,M and R_1,W, respectively.

Fig.1: a) Pulse diagram of SPECIAL with the HS (blue), GOIA (orange), and WURST (green) pulse; b) Scan scheme (exemplary): 1. day 1. session SPECIAL with HS, GOIA, and WURST; After repositioning the volunteer (1_2), scans were repeated as in 1_1. 2. day scans (i.e., one week later) 1. session, HS, and GOIA twice without repositioning. 2. session 2. day, WURST-SPECIAL twice without repositioning. c) Exemplary voxel position in the posterior cingulate cortex (PCC) of a volunteer.

Figure 1. Experimental setup: a) electrode placement and stimulator modes; b) setup for applying tDCS inside the MR scanner; and c) 45 minutes of tDCS-MRS scanning protocol timing diagram

Figure 3. Short term and Cumulative effect of GABA and Glx with respect to Creatine changes of Anode and Cathode

Figure 2. The paradigm of the visual stimulation: 3 s - a checkerboard flickering with a frequency of 4 Hz and 21 s - dark screen; scheme of fMRI and 1H MRS data acquisition during stimulus presentation.

Figure 1.T1-weighted image slices showing the voxel position for the single voxel spectroscopy in the activated visual cortex

Figure 1: Boxplot diagram for mIns/tNAA, mIns/tCr, tNAA/tCr and tCho/tCr and the different lesion categories “NWM”, “NAWM”, “nonBH”, “BH”, and “MRSI hotspot”. No differences between NWM and NAWM for each of the metabolites. Highly significant differences between lesion categories especially for mIns/tNAA

Figure 2: Conventional MRI with T1-weighted MP2RAGE and T2-weighted FLAIR. Yellow arrows point to “black hole” lesions which are clearly visible on mIns/tNAA, mIns/tCr and tNAA/tCr. The red arrow depicts a “non black hole” lesion, which is already apparent on mIns/tNAA and mIns/tCr, though not yet visible on FLAIR

Figure 1: T2-FLAIR, SWI and overlaid FLAIR-SWI images of(a) lesions categorized as “transition” with yellow arrow pointing towards an example, (b) “area” iron deposition indicated by a red arrow and (c) “rim” lesion with an abundant iron rim indicated by a green arrow.

Figure 2: Boxplot diagram for mIns/tNAA, mIns/tCr, tNAA/tCr and tCho/tCr and the different lesion categories “no iron”, “area”, “transition” and “rim”. Significant results were found between lesion categories for mIns/tNAA and tNAA/tCr, but mIns/tCr stays constantly elevated for all “black hole” lesions.

Fig. 1. Maps of GM, WM, and CSF partial volumes of a single subject. (A) High resolution reconstructed images obtained from 60x60 matrices of grey matter, white matter, and cerebrospinal fluid partial volume fractions. (B) Partial volume maps with 16x16 grid overlay composed of 10x10x10 mm voxels. (C) Lower resolution reconstructed images from 8x8 matrices of the three partial fractions. Central 8x8 grids are in bold. A color scale of blue to yellow is used, with blue being lower and yellow being greater partial volume fraction at each voxel. Partial fractions were obtained via VDI.

Fig. 2. Scatterplots of metabolite ratios to creatine (y-axis) against partial volume fractions of GM or WM (x-axis). GM is on the top row and WM is on the bottom row. The plots include a linear fit and report R-squared values. The intercepts of the linear fit where partial fractions are equal to 0.0 and 1.0 are shown by red markers (GM_0% and GM_100% for GM, and WM_0% and WM_100% for WM). Metabolites include tCho, tNAA, Glx, and mI. High correlations are observed with metabolites that are known to have different concentrations in grey and white matter.