Figure 4: Prospectively acquired in-vivo data comparing (a) GRAPPA reconstructions of standard HASTE and the proposed VFA scheme. VFA reduces blurring and retains 83% SNR in comparison to standard HASTE. (b) HASTE reconstructed with GRAPPA and POCS partial Fourier, and the proposed VFA reconstructed with SENSE-LORAKS using the external calibration scan. The reference scan allows prospective calibration of SENSE-LORAKS, and the proposed VFA regime acquired the entire stack of slices 2.3x faster than the standard HASTE.

Figure 5: Comparison between slice-by-slice GRAPPA, retrospective GRAPPA SMS, wave-encoded SMS and wave-LORAKS SMS. While slice-GRAPPA incurs some noise amplification, wave-LORAKS SMS produces cleaner images. Accounting for SAR associated with multi-band pulses in a prospective implementation, combining the proposed VFA technique with SMS could yield between 3-5x reduction in total scan time.

Representative reconstructions from the proposed motion correction strategy for 3D MRI of the fetus with isotropic millimetric spatial resolution. The top row shows the originally acquired (motion-blurred) data in approximate axial, sagittal, and coronal reformats while the bottom row shows substantial improvement in image quality and delineation of fetal brain structures after applying retrospective motion correction to the same data.

Animated figure depicting motion-resolved reconstructions of free-running 3D fetal MRI data. A-C) A low spatial resolution real-time image series with 500 ms temporal resolution demonstrates maternal respiration and gross fetal movement in three perpendicular planes. D-F) High spatial resolution images created from unique motion states identified in A) shown in the same views centered on the brain, allowing for co-registration to “stabilize” anatomical features of interest (G-I).

Figure 3. 4D flow exam of the fetal heart based on DUS fetal cardiac gating. Image examples with streamline visualization were selected from a case at 35 gestational weeks on 3.0T. Details see text.

DA = ductus arteriosus; PA = pulmonary artery; AoD = descending aorta; AoA = ascending aorta; RH = right heart; LH = left heart

Figure 2. Functional cine exams of the fetal heart based on DUS fetal cardiac gating on 3.0T. (A) Selected images in 4-chamber view (4CH) at end systole (ES) and end diastole (ED). (B) Selected images from three adjacent short-axis slices (SAX) from ES to ED. Arrows indicated the fetal heart at ES and ED.

Figure 1: Global and Regional Ecc in Healthy Boys and LGE(-) boys with DMD. Septal Ecc was significantly lower than anterior, lateral, and inferior Ecc and the lateral Ecc was significantly higher than septal, anterior, and inferior Ecc for both the controls or LGE(-) DMD patients. Furthermore, the LGE(-) DMD patients had significantly reduced septal Ecc compared to healthy volunteers. * p-value ≤ 0.05 is significant for a comparison within either controls or LGE(-) boys with DMD. # p-value ≤ 0.05 is significant for a comparison between controls and LGE(-) patients with DMD.

Figure 2: Receiver Operator Characteristic (ROC) curves for septal Ecc, lateral LV pre-contrast T1, and LVEF from a binomial logistic regression classifier for distinguishing between LGE(-) boys with DMD from healthy boys. Larger area under the curve (AUC) values indicate better classifier performance. Regarding differentiating LGE(-) boys with DMD from healthy boys, septal Ecc outperforms each individual biomarker and the combinaiton of septal Ecc, lateral LV pre-contrast T1, and LVEF outperforms the other biomarker combinations.

Figure 2. Conventional anatomical images and APT-weighted images within initial ROI (third column) and ROI after shrinkage with a cutoff value of 30% (fourth column), which were fused with an image without saturation. Columns 5 and 6 show quantitative T1 and T2 maps calculated from MIX images within ROI after shrinkage, respectively. Images were from a patient with low-risk neuroblastoma (a) and a patient with high-risk neuroblastoma (b).

Table 2. The ROC analysis for differentiation between low- and high-risk NBs

Figure 1: Schematic pulse sequence diagrams. (A) SC_2D_4BH: The polarity of MEGs is reversed (blue and yellow) every RF excitation. The mechanical wave polarity (green) stays the same across RF excitations. Each RF excitation triggers 3 motion cycles. (B) 3D_5sl_SBH : The polarity of MEGs remains the same (blue) across RF excitations. The mechanical wave polarity inverts (green and red) every RF excitation. Every 4th RF excitation triggers 7 motion cycles. Each sequence starts with dummy excitation of 1 s that ensures both acoustic and magnetization steady-state.

Figure 3: Comparison of liver shear stiffness values in individual slices for SC_2D_4BH and 3D_5sl_SBH techniques by Linear Regression and Bland-Altman Analysis. Values are based on manual analysis informed by a 95% confidence mask with matching ROIs on images from both techniques and maximal possible ROI in each. SC_2D_4BH: standard of care two-dimensional 4 slices through mid-liver with 13 second breath-hold per slice; 3D_5sl_SBH: 3D wave polarity-inversion motion encoding plus compressed SENSE acquisition in single breath-hold of <16s.

Figure 1: The anatomical locations for the 3 brain cortical regions that showed significant differences (FDR corrected P < 0.05) in mean cortical thickness between infants born to normal weight vs. obese pregnant women: a) the left pars opercularis (LPO) gyrus; b) the left pars triangularis (LPT) gyrus; and c) the left rostral middle frontal (LRMF) gyrus. The comparison of mean cortical thickness for these 3 regions between these two groups of infants is illustrated in d).

Figure 2: Partial correlation analysis showed significant negative correlations (P < 0.05) between maternal body fat mass percentage measured at ~12 weeks of pregnancy and infant brain mean cortical thickness measured at ~2 weeks of postnatal age in the a) left pars opercularis (LPO) gyrus; b) left pars triangularis (LPT) gyrus; and c) left rostral middle frontal (LRMF) gyrus.

Table 1. Demographic Information of Study Participants. SD = Standard deviation. Body mass index (BMI), full segmentation pancreatic proton density fat fraction (PDFF) and pancreatic volume were significantly higher in children with nonalcohol fatty liver disease (NAFLD) than healthy children.

Figure 4. Box-Whisker Plots of Pancreatic Proton Density Fat Fraction (pPDFF) from Different Methods. Full segmentation (FS) pPDFF was higher in children with nonalcoholic fatty liver disease (NAFLD) compared to healthy children (p < 0.001). The 3-ROI method underestimated pPDFF compared to FS (p < 0.001 for healthy and NAFLD). The superior slices in children with NAFLD had higher pPDFF than inferior slices (p = 0.003).

Figure 2. Path diagrams showing the indirect relationship between Cerebral Oxygen Delivery and Cognitive Composite Score mediated by (A) left thalamus (B) right thalamus (C) left caudate nucleus (D) right caudate nucleus. Standardised regression coefficients are reported; numbers in brackets show p-value *p<0.05 **p<0.01. ACME, Average Causal Mediation Effect; IMD, Index of Multiple Deprivation (a measure of socioeconomic status).

Figure 1. Volumetric brain development in neonates with CHD. Shaded areas represent ±1,2 and 3 standard deviations from the normative model mean, separately for female and male infants. Total Tissue Volume, TTV; Grey Matter, GM; White Matter, WM; Right, R; Left, L.

Figure 1. The schematic pipeline of construction of connectome trajectory (in pink) and its association (in green) with gene expression (in blue) from the 3^rd trimester to 8 years of age. Abbreviations: PLS = partial least squares; MFC = medial prefrontal cortex; V1C = primary visual cortex.

Figure 4. Gene enrichment analysis in top 10% genes associated with degree centrality demonstrated dynamic enrichment in (A) different cell types and (B) gene ontology terms for biological process through the 3^rd trimester and early childhood. Significance is labeled by color (red dot: p<0.05; gray dot: p≥0.05, FDR corrected). Abbreviations: Ex=excitatory; In=inhibitory neuron; Other=other types of cells in brain. Ex2b and Astro1 are subtypes of excitatory neuron and astrocyte, respectively.

Figure 1. Visual representation of the delineated white matter tracts in the neonatal atlas space. Shown in superior (left), anterior (centre) and lateral (right) views.

Figure 3. Multimodal general factors. (A) PCA variable contribution plot; the colours represent the contribution of the dMRI metric to the components. (B) Boxplots of the multimodal g-factors showing differences between the term and preterm group. Note that each participant is represented by 16 tracts. Reported statistics are standardised β for the PC1 or PC2 from linear mixed effect regression models and FDR-corrected p-values.

Design of our learning algorithm. The LR volumes are decomposed onto HR 2D slice stacks in the direction of slice selection, and their gradients are used as the output of the convolutional neural network (CNN) in the learning. On the other hand, all LR volumes are interpolated and combined into a blurred HR volume that is then resampled onto the LR image spaces to obtain the LR 2D slice stacks as the network input.

Reconstructed slices from a representative subject by our approach and the baselines on the clinical dataset. (a) SRCNN. (b) GGR. (c) Our approach (deepGG). Our approach achieved the highest quality in terms of image contrast and sharpness. The fine structures delineated by our approach are highlighted by the red arrows, in particular from the hippocampus in the coronal plane.

Figure 1. Representative in vivo proton MR spectra for the deep gray matter obtained from a neonate with hypo-ischemic encephalopathy: a) without and b) with water presaturation pulse in PRESS sequence (TE/TR = 30/5000 ms). Scales of a) and b) are not proportional. c) White voxel on the images show MR spectroscopy volumes of interest. tNAA = N-acetylaspartate and N-acetylaspartylglutamate.

Figure 2. Calibration curve obtained from NAA aqueous solution.

Figure 3: Subcortical and cortical volumes (top) and cortical surface area (bottom) for regions that were negatively associated with early diagnosis of cerebral palsy are shown in blue (shade is arbitrary). Grey/white regions are not significant.

Figure 1: Alignment (3rd column) of the 90-region Automated Anatomical Labeling atlas (2nd column) with diffusion space tractography (1st column) for an example subject. From here, we calculate the mean fractional anisotropy for all streamlines connecting each regional pair as a 90x90 symmetrical matrix.

Figure 1. Plots of correlation of 24h, 48h and average CBF with Cell death in BGT. (CBF = cerebral blood flow, BGT = basal ganglia thalamus)

Figure 2. Plots of correlation of 24h, 48h and average CBF with Iba1 ramification index in the cortex. Iba1 = ionized calcium binding adaptor molecule 1 (CBF = cerebral blood flow)

Figure 1. Regions with significant left/right difference after multiple comparison correction, based on MD (A) and FA (B). LIs of all 63 regions were plotted on the left (A1 and B1), with the positive values indicating leftward asymmetry and negative values indicating rightward asymmetry. *p < 0.05 by pairwise t-test. Regions with significant asymmetry were overlaid on the corresponding MD and FA maps in sagittal and axial views, with the color bar indicating the LI. Positive values (red) represented leftward asymmetry, and negative values (blue) represented rightward asymmetry.

Figure 4. FBA revealed extensive WM regions with significant asymmetry in terms of FDC. The three rows from top to bottom show results for TEA-1month, 1-3 months, 3-6 months, respectively. The color bar indicates the LI, with positive values (red) represent leftward asymmetry, and negative values (blue) represent rightward asymmetry. A clear inside-to-outside developmental change can be identified, with the lateralization primarily localizes in the central brain and major WM at TEA, but extend to peripheral and subcortical WM at 1-3 months which further increased at 3-6 months.

Figure 1. Multi-Site White Matter FA Model Without and With Consideration of Quality Control Measures. QC measures accounted for 47% of the remaining variance in FAWM after adjusting for PMA. Note the decrease in MSE and increase in the remaining variance in FA_WM explained by PMA after adjusting for QC measures.

Table 2. Single Site Association Between Whole White Matter FA and Postmenstrual Age at Scan Modeled Without and With Controlling for Quality Control Measures. All sites demonstrated marked improvement in model performance when controlling for simple QC measures (mean FD, SNR, CNR). Improved performance is indicated by increased partial R-square values, t-scores, and increased consistency in age-FA slope estimates across sites. All age-FA associations were significant at a p<10^-3 threshold for all models and sites.

Figure 2. Diffusion kurtosis and NODDI parameters from the post-mortem infant and an age-matched in vivo infant. The mean diffusivity is expressed in μm²/ms.

Figure 3. Left: Cortical radiality (dot product of principal diffusion direction and surface normal vector) of post mortem infant and an age-matched in vivo infant, calculated at mid-thickness. At the fundus of the central sulcus (black box), the post mortem data have high radiality but the in vivo data have low radiality. Right: principal diffusion directions for an axial plane containing central sulcus, the dashed white line indicates the grey white matter boundary. The principal diffusion directions at the fundus are radial in the post-mortem data but tangential in the in vivo data.

A 15-year-old boy with medulloblastoma in cerebellum (WHO grade Ⅳ). The lesion shows hyperintense on T2-weighted image(a), hypointense on T1-weighted image (b), and enhancement on post-contrast T1-weighted image (c). The lesion (ROI) demonstrates hypointense on the ADC map (d), D map (e), and Dk map (h), and hyperintense on the D* map (f), f map (g), and K map (i). The pathologic diagnosis was medulloblastoma with high cellularity of 4928(cell/mm2)(j), high Ki-67 index of 80% (k), and MVD of 1.4% (l) (magnification, × 200).

A 5-year-old boy with diffuse astrocytoma in brainstem (WHO grade II). The lesion shows hyperintense on T2-weighted image(a), hypointense on T1-weighted image (b), and enhancement on post-contrast T1-weighted image (c). The lesion (ROI) demonstrates hyperintense on the ADC map (d), D map (e), and Dk map (h), and hypointense on the D* map (f), f map (g), and K map (i). The pathologic diagnosis was diffuse astrocytoma with high cellularity of 1918(cell/mm2) (j), high Ki-67 index of 1.1% (k), and MVD of 0.9% (l) (magnification, × 200).

Figure 1. Averaged QSM maps for neonates and infants with chronological age (CA) of 0-150 days, 151-499 days and 500-750 days. Magnetic susceptibility of the basal ganglia and thalamus increases with age.

Figure 2. Scattered plots of magnetic susceptibility of the basal ganglia and chronological age. Magnetic susceptibility of the CN, GP and PT correlates with chronological age (CA).

Fig. 1 spatiotemporal_MD_GM

Fig. 2 spatiotemporal_MD_sWM

Figure 1: Cortical microstructure processing and analysis framework. A) T1-weighted data corrected for bias field and co-registered to an upsampled b=0 s/mm² (1x1x1mm); B) T1-weighted data processed using Freesurfer[13,24], registered to MNI space to obtain 7 functionally defined networks[12], and resampled to dMRI space to obtain network labels intersecting the cortical ribbon; C) representative maps of microstructural measures; D) high repeatability of SANDI measures at 300mT/m[9]; E) developmental patterns of macro- and microstructure averaged over the cortical ribbon

Figure 2: Relationship between cortical microstructure and age, grouped by measure and coloured by network type. Abbreviations: CTh = cortical thickness, in mm; f = signal fraction; MD = mean diffusivity, in 10^-3 mm²/s; OD = orientation dispersion; R = apparent radius, in µm. Note: MD was estimated using b=0,6000 s/mm² to improve sensitisation to cortical microarchitecture[17]

Figure 1. Canonical correlations (A) and loadings (B-D) for each analysis. Colors show the loading strength of phenotype (B) and white matter (C, D) measures on their respective latent variates for each canonical correlation analysis. Warmer colors indicate stronger positive loadings (higher values on an individual measure); cooler colors indicate stronger negative loadings (lower values on an individual measure). For clarity, loadings between -0.2 and 0.2 are shown in white.

Figure 2 A. Boxplot for differences in discriminant function between ADHD and TDC. (p<0.001) B. Table shows CCA constructs for each analysis, the effect they represent, and the standardized coefficients resulted from the linear discriminant analysis.

Figure 1: Example voxel placement in the primary sensorimotor cortex (SM1) and thalamus (THAL), along with respective data. Spectra are color-coded by acquisition phase: gray = macromolecule-suppressed MEGA-PRESS (Phase 1); blue = macromolecule-suppressed MEGA-PRESS with prospective frequency correction (Phase 2); red = HERMES (Phase 3).

Figure 2: Group comparison of tissue-corrected metabolite levels (residuals after linear-mixed effects model accounting for age, sex, and acquisition phase) between children with ASD and typically developing children. Sensorimotor Glx levels are significantly higher in children with ASD (a), while no significant group differences were found for thalamus Glx (b), sensorimotor GABA (c), or thalamus GABA (d).

Figure 2: The average amygdala functional connectivity across all subjects coloured by z-statistic.

Figure 5: The relationship between the PROMIS maternal anxiety measures and infant amygdala-prefrontal functional connectivity.

Geometric mean $$$R_2$$$ as a function of fiber angle in newborns and best fits for the dipole-dipole model, the Knight model and the diffusion model.

Predictions for the $$$R_2$$$ orientation dependence for the dipole-dipole model, the Knight model and the diffusion model.

Figure 3. Bilaterally segmented sensorimotor tracts. (A) corticospinal tract in coronal view; (B) superior thalamic radiations (motor) in coronal view; (C) superior thalamic radiations (sensory) in coronal view; (D) posterior thalamic radiations in axial view

*All fibers of the STRS are not visible in coronal view, as the tract continues from the thalamus to the post-central gyrus

Figure 2. Location of regions of interest (ROIs) on the group fixel plot and the segmented right corticospinal tract. (A) Seed point ROI covering the right cerebral peduncle on an axial view; (B) Waypoint ROI covering the right posterior limb of the internal capsule on an axial view; (C) Waypoint ROI covering the right precentral gyrus on an axial view; (D-F) Right corticospinal tract on sagittal, coronal and axial views, respectively.

Figure 1. Example neonatal MTsat and MTR parametric maps.

Table 1. Participant demographics.

Figure 1. Representative functionally-related and specific IC maps overlaid on the study template.

Figure 2. Overlay of the compiled core, fine and gross motor, expressive and receptive language, and visual receptive network masks.

Figure 2: Group-level large-scale brain networks derived from ICA analysis (a; orange = positive correlation, blue = negative correlation; FPN = frontoparietal network, SN= salience network, DMN = default mode network, DAN = dorsal attention network) and the methods for extracting local rsfMRI variability (b), cerebroarterial density (c) and cerebral microbleed (CMB) burden (d).

Figure 4: Relationship between imaging parameters and verbal memory performance (dependent variable). DMN var = default mode network variability, vvol= normalized vessel volume.

Figure 2. Overview of semi-supervised graph convolutional network to predict motor impairments in very preterm infants. The input is a cohort graph that describes the inter-subject similarities among training (both labeled and unlabeled) and testing (subjects-to-classify) subjects. The first graph convolutional layer contains 16 graph convolutional filters. The second graph convolutional layer contains 8 graph convolutional filters. The number of graph convolutional filters were selected from empirical values [8, 16, 32, 64] based on validation performance.

Figure 1. (A) Cohort graph describes the inter-subject similarities among the training (both labeled and unlabeled) and testing (subjects to classify) subjects/nodes. (B) Structural connectome feature vectors are used as the node features in the cohort graph. (C) Diffuse white matter abnormality volumes and global brain abnormality scores are used to calculate the inter-subject similarity as edge weights in the cohort graph.

Fig.2 Clinical-imaging radiomics nomogram

Fig. 4 ROC curves of each model in the training set and validation set

ANN and CF derived MWF maps in children.

Linear regression between the ANN-MWF and CF-MWF values.

Figure 1. Selected examples of the pediatric neuro-onco MRI exams in the current study comparing C-SENSE and conventional scans.

Table 2. Spatial resolution improvement and scan duration reduction by C-SENSE in the pediatric neuro-onco MRI pulse sequences in comparison to the conventional scans. Details see text.

ACQ = Acquired; SENSE = SENSitivity Encoding; C-SENSE = Compressed SENSE; TFE = Turbo field echo; FLAIR = fluid attenuated inversion recovery; TSE = Turbo spin echo.

Figure 2. Representative images shown for (A) a healthy child (H₁) and (B) a child with liver disease (L₄). The coronal reformats show the chosen slices. The 2-D TE1 image and PDFF maps were inputs to the network. SAT (white) and VAT (gray) manual annotations are displayed. Both networks yielded comparable Dice-SAT. The proposed network yielded higher Dice-VAT than U-Net for H₁ and L₄, with differences shown in zoomed insets. The U-Net misclassified VAT as seen in violet boxes for H₁ on chosen slice 2, and for L₄ on chosen slices 1 and 2. Underlined Dice scores represent the higher score.

Figure 1. The structure of the proposed network. (A) ResPath Dense U-Net takes 2-D TE₁ images and PDFF maps as inputs, and outputs segmentation masks of 3 classes: SAT, VAT, and the others. (B) Dense connections, shown inside a dense block, improve the generalizability of the features and overcomes the vanishing gradient problem. (C) ResPath skip connections mitigate the semantic gaps during the concatenation of encoder and decoder features.

LAEQ of Scans Organized by Scanner. The LAEQ ranges on each scanner vary between 8.49 - 11.9 dBA, indicating that choices in scan type can make a difference. Scanner D, the neonatal scanner, is significantly quieter than other machines. Regardless of scan type, every scan on Scanner G is louder than most other scans.

LAEQ of neonatal protocols by scanner relative to the IEC MRI hearing safety threshold. Compared to the IEC MRI hearing safety limit (red dots), every neonatal protocol LAEQ (blue stars) with hearing protection (blue spotted area) is below the limit. The maximum hearing protection limit was calculated using an OSHA standard ¹⁴.

Figure 1. Parameters for the three different MPRAGE sequences used in this study.

Figure 2. (A) Representative images from MPRPAGE, wave-CAIPI MPRAGE, and vNav Moco MEMPRAGE sequences from the same patient. The patient was injected with contrast enhancement for the scan. (B) Contrast-to-noise ratio (CNR) for five different cortical areas. The left figure is from the cohort that did not receive any contract enhancement, and the right figure is from the cohort that did. The analysis was performed using the three different sequences.

Schematic representation of through plane motion and refocusing pulse modifications. (A) shows a coronal acquisition during expiration and inspiration, where abdominal contents are displaced by the diaphragm. (B) illustrates the proposed refocusing pulse modification by doubling its thickness to mitigate long T2 signal loss.

Comparison of respiratory triggered, free-breathing, and modified 2x refocusing pulse free breathing T2-weighted fat suppressed multi-shot TSE (BLADE). Arrow shows severe signal loss for the standard free breathing sequence that is mitigated in the modified 2x refocusing pulse sequence. Acquisition times (AT) for each sequence are displayed in mm:ss.

Figure 4 – A) and B) 3D render of the structural elements from Fig. 1A-D. A infant model¹¹ was added for better visualization of the positioning. The Tx-array coil holder was slightly moved out of the final position. C) 3D render of the structural elements from Fig. 1A-C. The fitting of the baby model can be clearly evaluated. The main support structure (Fig. 1E) is not shown to simplify the view.

Figure 1 - 3D render of the main parts of the mechanical frame for infants at 7T, with A) the Tx-array coil holder showing the position of the dipole elements, B) the bed structure, C) the posterior Rx-array frame, D) the anterior Rx-array frame and E) the main support structure. The parts displayed in light blue (in C and E) correspond to the Teflon pads used for displacement between the brain-isocentred and heart-isocentred positions.

Fig 1. The hoods. 1a: Prototype that has been used this past year. 1b: Mechanical design of the new hood. 1c: Design of easy access point to the neonate. 1d: the finished product.

Fig 3. Test results. a: Results of calibrated tests in the acoustic room. b: Results of tests inside the MRI during various scan protocols.

Figure 1. Correlation plots of T1 hepatic values using 2D-MOLLI versus 3D-QALAS (A) mean T1 and (B) T1 of individual slices

Figure 2. Bland-Altman plots of T1 hepatic values using 2D-MOLLI v/s 3D-QALAS (A) mean T1 and (B) T1 of individual slices.

Fig.1 Setting ROI and FWM on DWI and ADC

Fig.2 Box plot of rADC values in HC, HBn and HBh groups

Figure 1: Schematic pulse sequence diagrams. (A) SC_2D_4BH: The polarity of MEGs is reversed (blue and yellow) every RF excitation. The mechanical wave polarity (green) stays the same across each RF excitation. Each RF excitation triggers 3 motion cycles. (B) 2D_CS_HBH with inflow saturation: The polarity of MEGs remains the same (blue) across RF excitations. The mechanical wave polarity inverts (green and red) every RF excitation. (C) 2D_CS_HBH no inflow saturation. Every other RF excitation triggers 4/3 motion cycles.

Figure 3: Comparison of liver shear stiffness values in individual slices for SC_2D_4BH and 2D_CS_HBH techniques by Linear Regression and Bland-Altman Analysis. Values are based on manual analysis informed by a 95% confidence mask with matching ROIs on images from both techniques and maximal possible ROI in each. SC_2D_4BH: standard of care two-dimensional 4 slices through mid-liver with 13 second breath-hold per slice; 2D_CS_HBH: two-dimensional polarity-inversion motion encoding plus compressed SENSE acquisition in half the SC breath-hold.

FIG 1 LIC ROIs

FIG 2 Graph comparison of LIC - all results

FIG 3 Bland-Altman plots

FIG 4 Sample individual results

FIG 5 ROC curves

Figure 2. Images reconstructed using hard-gating and 3 soft-gating weighting functions (exponential, inverse and linear). aSNR for each reconstructed image using the different weighting function is based on the optimized parameter for each weighting function. Area in the red rectangle encompasses the superior-inferior lines used to generate linearly scaled mean signal intensity profiles at the lung-diaphragm boundary in figure 3.

Figure 1. (A) A smoothed waveform fit through initial phase of free induction decay (FID) with data binned into 8 different bins from 0 to 7. (B) Bin assignment of the data based on the median wave in (A), where bin 8 represents bulk-motion and is discarded. (C) Hard-gating and three soft-gating weighting functions (exponential, inverse and linear) shown here with respect to Bin 0.

Figure 1. Example of tractography. Example of tractography of two children, ages 12 (boy) and 15 (girl).

Figure 1. Bland-Altman plots of predicted and measured VH and THG. Bland-Altman plots showed reduced error in DTI-P/M predictions (top) compared to bone-age predictions (bottom). Also, DTI-P/M predictions did not have bias (zero difference within the 95%-confidence interval), while bone-age predictions had a significant bias. The dashed line represents the mean difference between predicted and measured; the dot-dash line the two-sigma range. The light gray area indicates the 95%-confidence interval of the mean difference. Purple circles are boys and dark pink girls.

Figure 1: Study protocol, MRI planning, and quantitative flow measurements obtained with PC-MRI (note: renal veins not shown).

Figure 2: Scatter plots of: (a) total left and right arterial renal blood flow, and (b) the average left and right peak renal blood flow.

Box plots diagram of fat fraction values of healthy children and patients with leukemia

Pearson correlation analysis results for fat fraction (FF) quantification by mDixon-Quant and MR-spectroscopy

Fig 1. 25-weeks’ gestation fetus with sacrococcygeal type IV teratoma (a) (b) Sagittal and coronal T2-weighted image shows cystic mass (arrow) arising from coccyx, with septa (small arrow) is seen (b). (c) Sagittal Susceptibility-weighted imaging (SWI) shows fetal sacrococcygeal vertebra were nicely demonstrated

Fig 4. 28-weeks’ gestation fetus with sacrococcygeal type IV teratoma(a) (b) Coronal and sagittal T2-weighted image shows solid mass (arrow) arising from sacrococcygeal region with unilateral hydronephrosis (white arrow)(c,d) Diffusion Weighted Imaging (DWI) and corresponding apparent diffusion coefficient (ADC) showed the mass diffusion limited

Figure 1. A. T2-HASTE axial images of the fetal liver in a mother with fetal growth restriction to provide an anatomical reference. B. Axial 3-D stack-of-radial FB-MRI PDFF map (range: 0-100%) of the fetal liver. Circular ROIs with 1-cm² area were used to measure the fetal liver PDFF. C. Axial 3-D stack-of-radial FB-MRI PDFF maps of the fetal abdomen. The red ROI was used to measure fetal abdominal subcutaneous adipose tissue (SAT) volume and PDFF. D. Coronal reformatted PDFF map of fetus. The green dashed and solid lines indicate the axial slices in panels B and C, respectively.

Figure 2. A. Axial 3-D stack-of-radial FB-MRI PDFF maps (range: 0-100%) of the maternal liver in a mother with gestational diabetes. Cyan outline measured liver PDFF. Purple ROI measured subcutaneous adipose tissue (SAT) PDFF and volume. B. Axial 3-D stack-of-radial FB-MRI PDFF maps (range: 0-100%) of the maternal abdomen. Red ROI measured visceral adipose tissue (VAT) PDFF and volume. C. Coronal reformatted PDFF map of the maternal liver and abdomen. The green dashed and solid lines indicate the axial slices in panels A and B, respectively.