Figure 1: Percentage of frequency of qualitative scores in 5-point scale (Excellent:5; Good:4; Fair:3; Poor:2 and Uninterpretable:1) for four criteria for qualitative evaluation of IVIM parametric maps evaluated by five IVIM analysis methodologies a) BE, b) BEseg-2, c)BEseg-1, d)BE+TV and e) BE+HPF.

Table 1: Qualitative scores of IVIM parametric maps evaluated by five IVIM analysis methodologies and Friedman test statistics for statistical significance (p<0.05). Scores are in mean± standard deviation.

Fig.5 Forest plots of ADC value between residual or recurrent HCCs and necrotic tumors.

Fig.4. Forest plots of ROC for DWI in the detection of residual or recurrent HCCs after TACE.

Figure 1. Mean whole-skeleton FA values obtained in the controls of three cohorts with the correction method used in the original ENIGMA processing with eddy_correct, compared with the three methods we tested: recomputed with eddy_correct, eddy, and TORTOISE. P-values were obtained from paired-t test analysis.

Table 1. Age of peak at onset decreased whole-brain FA value on trajectory from the original publication and the TORTOISE DT registration to DT study-specific template in SCZ patients and controls in each cohort.

Figure 2. Synthetic phantom correction calculations using FA measurements from DTI imaging of the synthetic phantom unidirectional fiber blocks. The first set of scans (A) were used to determine the scanner specific correction factors which were then applied to the second set of scans.For all scanners (B). After the correction (C), a reduction in cross scanner variability is observed.

Figure 3. Along-tract FA analysis of an individual human phantom across multiple sites. The tracts presented are CC Body (A), right CST (B), Splenium (C), and left SLF (D). Present to varying degrees in all the tracts are regions that are more sensitive to scanner variability (blue arrows) and regions that are more reliably measured (red arrows) regardless of MRI used.

F-stats from 2x2 ANOVA for gFA and FW metrics. In panel (a) the significant clusters for the main effect of TP are shown (for only gFA). Panels (b) and (c) show the significant clusters for group×TP interaction (FW and gFA, respectively). Table in panel (d) shows the summary of the results.

F-stats from 2x2 ANOVA for VBM analysis. In panel (a) the significant clusters for the main effect of TP are shown, while panel (b) shows the significant clusters for group×TP interaction. Table in panel (c) shows the summary of the results.

FA and kFA for ROIs with significant group differences.

MD and MK for ROIs with significant group differences.

Radiomics workflow

Study flowchart

Figure 1: b750 images and ADC maps of liver metastases from two patients from each cohort: a 53-year old woman (top / cohort 1) with colorectal cancer and a 61-year old man with bowel cancer (bottom / cohort 2).Note the increased homogeneity signal across the slice on both b-value and ADC images when using the prototype DWI with advanced options (columns 1 vs. 4). Moreover, the advanced option allows for a better delineation of the liver or blood vessels as seen on the b750 images.

Table 2: Mean scores of the overall image quality for each type of images (b100, b750 and ADC map) and each DWI method as derived from the 25-patient cohort. Highlighted in green are the top scores and in red the lowest scores given by the two radiologists (scoring scale was from 1 to 3).

Figure 2. (a) Mean ADC amplitudes from the first to the seventh harmonic in the frontal white matter in the iNPH and control groups and (b) representative images of the mean ADC amplitude.

Figure 4. (a) ADC amplitudes in each frequency component in the frontal white matter in the iNPH and control groups and (b-d) representative images in each frequency component.

Predicted and observed tract - task correlations

Results from exploratory analysis showing relationships between left uncinate HMOA, age and planning performance

Figure 1: TBSS partial correlation for α and blood pressure exposure over 30 years for 77 longitudinal CARDIA participants (57.0±3.4 years). Red represents significantly changed voxels where there is a significant negative association between α and blood pressure exposure (r=-0.362, p=0.032). Cool map represents the white matter hyperintensity (WMH) probability map from FLAIR. Statistical analyses were adjusted for age, education, race, smoking history, diabetes, and cholesterol risks as covariates and corrected for multiple comparisons using permutation-based testing.

Table 1: Demographics, characteristics, clinical measures, and risk factors for the year 30 of the CARDIA cohort.

Figure 1 Lesion (green) and probability (red) maps of selected (Anterior thalamic radiation and forceps major) reconstructed tracts overlaid on a FLAIR image for the three tractography configurations: using b=700, b=2000 or both.

Figure 2 Example of diffusion metrics Z-Score (left axis) and lesion average probability (right axis) profile along a specific tract for DTI, DKI and NODDI measures. Position 0 indicates the centre of mass of the tract probabilistic map. Pearson correlations are indicated for each plotted metric

Figure 3. Local alterations in betweenness centrality of the structural brain connectivity networks. Statistical significance is illustrated by the color of the node from white (not significant) via yellow and red to black (most significant) as shown by the color bar. The size of the node reflects the volume of the gray matter area. The color of the edges corresponds to the direction – red: left (L)-right (R), blue: inferior (I)-superior (S), green: anterior (A)-posterior (P) The betweenness centrality of the left angular gyrus was significantly decreased (P=0.00027).

Figure 2. Local microstructural differences in mean diffusivity (A, B, E) and axial diffusivity (C, D). In Fig. 2E the location of the left angular gyrus is highlighted. The color scale from red to yellow describes the statistical significance of the decrease in the microstructural metric in the loss of consciousness (LOC) state compared to awake state. The white matter skeleton is visualized in green. The images are presented in radiological convention (left hemisphere on the right and vice versa).

Figure2: Top row: coronal view of the acute postsurgical T2WI from four patients where final target has been identified (blue dot), based on the criteria described in figure1. Bottom row: coronal view of the high quality presurgical DTI and the corresponding location of the final target, for each patient.

Figure3: Left column: In patient 3, the location of the first target point of sonification on postsurgical T2WI and its corresponding location on the high quality DTI. The sub threshold initial sonication at this site resulted in patient experiencing slight paresthesia. Right column: The final target location selected for surgery based on the positive functional feedback received from the patient and its corresponding location on the presurgical high quality DTI.

Figure 1. Probabilistic tractography of the AI tracts.

Figure 2. The differences of the diffusion findings of insula tracts between smokers and nonsmokers. Compared with nonsmokers, in the left hemisphere, smokers showed lower FA values of PI-NAc and AI-NAc. In the right hemisphere, smokers showed increased FA values of AI-mOFC, PI-mOFC and PI-NAc; decreased AD values of AI-mOFC, PI-mOFC and PI-NAc; decreased RD values of AI-mOFC, PI-mOFC, and PI-NAc; as well as decreased MD values of AI-mOFC, PI-mOFC, and PI-NAc.

Figure 1. Rich club coefficient of group-averaged structural brain networks at (A) within 1 week, and (B) 1, (C) 3 and (D) 6 months after first-time ischemic acute stroke. * p < 0.01 after FDR correction, indicating a significant rich-club organization at nodal degree k.

Table 1. Rich-club nodes at different time points after stroke

Figure 1: Cortical streamlines for STN and MSR: A. for a HCP subject, and C. for a PD patient. Histograms for HCP dataset (B.) and for PD dataset (D.) showing averaged connectivity of different cortical areas to the STN and MSR. The percentages of mean streamline numbers connecting the different sensorimotor (green), associative (blue) and limbic (red) cortical areas to the STN (dark bars) and to the MSR (bright bars) are shown for the left hemisphere (LH).

Figure 2: 3D localization of the three cortical clusters in both the STN and MSR in native spaces A. for 6 HCP subjects, and B. for 6 PD patients

Fig. 1 All models of diffusion MR imaging. This image shows one patient with high-grade brain tumor, where the first image is DWI at b = 1500s/mm²and from image a) to p) represents "ADCmap", "Dmap", "D_f map", "Dsmap" of IVIM ,"DDCmap","$$$\alpha$$$map" of SEM,"Dmap","$$$\mu$$$map","$$$\beta_f$$$map" of FROC,"Dcmap","$$$\alpha_c$$$map","$$$\beta_c$$$map" of CTRW, "D_k map","Kmap"of DKI and "ADC_Smap ” of SM respectively.

Fig. 2 Jointly parameters in each models and this picture shows ROC analysis of 7 models.

Figure 1. Anatomical distribution of logistic regression voxel-wise weights (i.e., odds) that contribute to correct classification, for the four best classifying dMRI measures.

Table 1. Classification accuracy and MCI recall for all tested measures. Recall values are multiples of 10% as there were 10 MCI individuals in the test group. As the training and test group sizes increase, these measures may differ more from one another.

Figure 2. Local decreases in A) strength and B) local efficiency of the structural brain connectivity networks in patients who did not survive after OHCA compared to the survivors. The color of the node indicates the statistical significance of the results. The results were adjusted for false discovery rate (FDR) with a significance threshold 0.05. Age, gender, and imaging site were used as covariates in the statistical model.

Figure 1. Global structural brain connectivity network differences between the survivors (group 0) and those who did not survive (group 1) after OHCA. The P-values were adjusted for false discovery rate (FDR). Age, gender, and imaging site were used as covariates in the statistical model.

Fig 2: CSF and GM density change in the opposite direction at the superior frontal and parietal part of the cerebrum, and along the ventricles and the Sylvian fissure between pre-flight and post-flight. These changes are then largely reversed seven months after spaceflight as revealed by the comparison of follow-up to post-flight. Results are scaled by effect size and are overlayed onto the group template image.

Fig 1: GM mass increases from pre- to post-flight are found in the basal ganglia. WM mass increased from pre- to post-flight in the cerebellum, in a part of the corticospinal tract and in the pre- and postcentral gyri. No significant decreases in GM or WM tissue mass were found between pre- and post-flight. No significant changes were observed between post-flight and follow-up, and between pre-flight and follow-up. Results are scaled by effect size and are overlayed onto the group template image.

Figure 1: Images of patient 2 (female, 41-year-old) with an RCC in coronal view. The upper row shows the images of T2-TSE, non-DW PSF-EPI, mean DW PSF-EPI, non-DW SS-EPI and mean DW SS-EPI from left to right. In PSF-EPI, resolution = 1 x 1 x 2 mm³ and b-value = 800 mm/s² were used. The bottom row shows the same images as those of upper but overlaid with the edges extracted from T2-TSE. Image distortion and signal piling up exist in SS-EPI (red arrow).

Figure 2: Images of patient 1 (female, 36-year-old) with microadenomas (8.3 mm) in coronal (upper) and sagittal view (bottom). From left to Right: T1-TSE, zoom-in T1-TSE, non-DW PSF-EPI, mean DWI(1 x 1 x 2 mm³, b=800 mm/s²) and ADC maps. The lesion shows hypointense in T1-TSE and hyperintense in distortion-free DWI (yellow arrows).

Figure 1. Age-related differences in mean diffusivity (MD). MD is significantly correlated with age across the cerebral cortex, particularly in the cingulate, insular and superior temporal cortices.

Figure 2. Age-related differences in cortical thickness (CT). CT is significantly correlated with age across the cerebral cortex, particularly in the motor cortex, although there are less significant regions with age-related differences in CT than there are in MD.

White and grey matter volume increase significantly at peak stress during hemodialysis while cerebrospinal fluid volume decreases (p<0.05). These changes are consistent with brain swelling and indicate ionic edema.

The colored regions show where each diffusion metric increased significantly (P<0.05) at peak stress during hemodialysis. The blue and yellow regions correspond to fractional anisotropy and axial diffusivity, respectively, and indicate cytotoxic edema. The green and red regions correspond to mean diffusivity. Radial diffusivity is shown in red.

Figure 1. Top row: Template fibre map showing regions included in analysis. Bottom row: Regions of significant correlation (FWE-corrected p<0.05) between fibre density and brain parenchymal fraction at baseline.

Figure 2. Regions showing significant longitudinal decline in fibre density over time in patients (FWE-corrected p<0.05).

Figure 1. White matter fiber cross-section (FC) increases in neuropathic pain. Statistically significant (FWE-corrected) FC increases in SCI subjects with neuropathic pain compared to healthy SCI controls were seen in the posterior splenium of the corpus callosum/retrosplenial complex region. Z values are relative to the AC/PC plane.

Figure 2. White matter fiber density and cross-section (FDC) increases in neuropathic pain. Statistically significant (FWE-corrected p<0.1) FDC increases in SCI subjects with neuropathic pain compared to healthy SCI controls were seen in the posterior splenium of the corpus callosum/ retrosplenial complex region. Z values are relative to the AC/PC plane.

Fig. 2. The white matter (WM) tracts with significant between-group differences in FA (A), MD (B) and LDH (C), respectively. “R” and “L” indicate right and left side, respectively. RUF, right uncinate fasciculus; RCST, right corticospinal tract; LCST, left corticospinal tract; LCH, left cingulum hippocampus; RSLF, right superior longitudinal fasciculus (temporal part); LATR, left anterior thalamic radiation; RATR, right anterior thalamic radiation; LIFOF, left inferior frontal-occipital fasciculus. These tracts were defined by the JHU ICBMDTI-81 white matter atlas.

Fig. 4. The results of receiver operating characteristic (ROC) curve analysis.

Figure 1: Brain regions with increased MD values in T2DM patients at baseline and 6 months follow-up appeared over controls. These sites included the (arrows) anterior and posterior cingulate (a-c), frontal (d) and prefrontal (e), insular cortices (f), and cerebellum (g). Few regions emerged with widespread damage in follow-up over baseline T2DM patients, and these sites (ellipses) included the caudate (h), middle cingulate (i), hypothalamus (j), and hippocampus (k). All images are in neurological convention (L = left; R = right). Color bar indicates t-statistic values.

Figure 2: Paired t-test exhibit higher MD values in 6-months follow-up T2DM patients over baseline. The regions with increased MD values included the cerebellum (a), mid (b) and posterior (d) cingulate, thalamus (c, e), insular cortices (f, j), caudate (g), amygdala (h), and hippocampus (i). Figure conventions are same as in Figure 1.

Voxel-based analysis for the F-statistics across groups for all analysis methods. VBM results are also shown uncorrected for family-wise error (FWE) in panel (a). For all other panels, we report clusters at a significance threshold of p-value < 0.05 corrected for FWE.

Voxel-based correlations between MoCA/Clock Draw/FAST and VBM/ADC/IVIM metrics. The plots, with the relative mean of R², represent the correlations of the average values inside the significant clusters.

Figure 1. The ADC distribution of cholesteatoma and non-cholesteatoma. The mean ADC values of these two sets of statistical measurements are: cholesteatoma: 860.77×10-6 mm2/s and non-cholesteatoma: 1529.78×10-6 mm2/s (Wilcoxon rank-sum test, p<0.01).

Figure 2. The area under the ROC curve of the mean ADC value is 0.98, which indicates that the ADC values have very good predictive ability for the detection and differentiation of cholesteatoma and non-cholesteatoma (granulation tissue or cholesterol granulation).

Study workflow illustrating the process of moving from native space images (three images on left), registering with the respective template for each method, applying the transform generated from that registration to the free water signal fraction map, and obtaining the free water signal fraction maps in stereotaxic MNI space (two images on right). This process was repeated 5 independent times for each of the 5 subjects.

The mean squared difference between each of the 5 independent registration attempts for each of the 5 subjects involved in analysis are presented, alongside the group mean (±SE). Grey bars represent the SyN intensity-based registration implemented in ANTs and yellow bars represent the WM-FOD based registration implemented in MRtrix.

Figure 1 The data proccessing flow diagram of DOSY

Figure 2. Reconstructed spectrum for experimental data QGC consisting of quinine (100MmM/L), geraniol (100mMMm/L), camphene (200 mMMm/L) dissolved in deuterated methanol (a) mono-exponential fitting, (b) SILT, and (c) our method named LRSpILT. Areas marked by red rectangular are with overlapping peaks. Artefacts in (b) and (c) are marked with black arrow. Highlighted areas are projected onto the diffusion coefficient dimension, and shown in Figure 4. The center diffusion coefficient values of components are marked with dashed lines.

Fig 2 Demonstration of progressive changes of DTI indices (FA, MD, AD, RD) in chimpanzee optic nerves.

Fig 3. progressive evolution of DTI indices of the optic nerve in chimpanzees with the ages of 13-20(n=13), 22-38(n=10), and 40-56 (n=7) years old. *, p<0.07; #, p<0.1, compared 22-38y group by using one-way ANOVA. Error bar: standard deviation

Figure 2: (a) Representative whole brain tractography demonstrated in 3 orthogonal imaging planes and whole brain automatic white matter segmentation results from volunteer #1 obtained at 0.5T (top) and 1.5T (bottom). (b) Analogous tractography images obtained from volunteer #2.

Figure 1: (a) Representative processed FA (left), ADC (middle), and RGB (right) images from volunteer #1 obtained at 0.5T (top) and 1.5T (bottom). (b) Analogous images from volunteer #2.

Figure 4 - Noise free simulations for two different sequences and their reconstructed images. a) Sample with four different tissues each with their respective relaxation times. The LS tissue (in white) has similar values to the WM. b) Partial EPI sequence in clinical MR scheme with b=1000 s/mm², B0=1.5T, sm=200T/m/s, gm=30mT/m, res=2mm. c) Spiral Sequence, single shot in LF with b=700 s/mm², B0=0.2T, sm=50T/m/s, gm =14mT/m, res=3mm. Images were reconstructed for both sequence with no DW (left), with DW (same color bar than no DW) (center) and ADC maps (right).

Figure 2 - SNR and CNR for varying B0 field intensity, gm, spatial resolution (displayed using a logarithmic scale) and b-values. a) Variation of SNR across gm values and B0 field, for a fixed res = 4mm. b) Variation of SNR across resolution values and B0 field, for a fixed gm = 20 mT/m. c) Variation of CNR across resolution values and B0 field, for a fixed gm = 20 mT/m. d) Variation of CNR across b-values, for a fixed gm = 20 mT/m and res = 4mm.

Fig. 3. Parameters maps generated from CTRW diffusion model for two image slices of an APP knock-in mouse and corresponding slices in WT control. A. α map with arrow pointed to CC; B, D maps (mm²/s) with arrow pointed to AC (upper panel) and CC (lower panel).

Fig. 4. Hippocampus showing lower values in α maps (lower panel) in APP knock-in mouse comparing to the WT control. Upper panel: diffusion-weighted images with b = 4050 s/mm², with gray bar showing the normalized signal intensity. Lower panel: α maps with color bar showing the extracted α values. Arrow points to the hippocampus.

ΔT along the CSF flow speed at cerebral aqueduct.

Bland-Altman plot between MRS and DWI based temperature.

Fig. 3 Generated T2w-FLAIR images using the proposed method with or without PSF-DWI (denoted by CGAN and CGAN w/o DWI, respectively) and their counterpart of the acquired T2w-FLAIR. SSIM values between the generated and acquired T2w-FLAIR were shown on each image. The input T2W-TSE, b = 0 s/mm² and mean DWI images acquired by PSF-EPI were also provided.

Fig. 2 Network architecture of the generator and the discriminator. Generator: a 2D U-net was designed with 19 convolutional layers (kernel size =3×3), 4 convolutional layers with strides for downsampling (kernel size = 2×2, strides = 2×1), 4 deconvolutional layers with strides for upsampling (kernel size = 2×2, strides = 2×2), and 4 feature contracting paths. Discriminator: 4 convolutional layers (kernel size =3×3) and 4 convolutional layers with strides for downsampling (kernel size = 3×3, strides = 2×2). Batch-normalization (BN) and ReLU were used for each layer.

Figure 4. Summary of the results obtained in the diverse analyses. Significant differences between episodic migraine (EM) and controls were found only with the AMURA return-to-origin probability (RTOP), detecting white matter alterations non-measurable with Diffusion Tensor Imaging (DTI). Differences between chronic migraine (CM) and EM were found with return-to-plane probability (RTPP) obtained with AMURA and axial diffusivity (AD) from DTI. No differences were found between CM and controls.

Figure 5. Visual comparison of Diffusion Tensor Imaging (DTI) and Apparent Measures Using Reduced Acquisitions (AMURA). The first row contains the DTI measures, from left to right: axial (AD), mean (MD) and radial diffusivity (RD). The second row contains AMURA metrics, from left to right: return-to-plane (RTPP), return-to-origin (RTOP) and return-to-axis probability (RTAP).

Figure 2. Evolution in the number of regions with significant differences using 61, 40 and 21 diffusion gradient directions. Axial diffusivity was compared between chronic and episodic migraine patients. A decrease in the number of gradients can be compensated by increasing sample size. Number of subjects per group for the diverse iterations is shown in the horizontal axis, corresponding the maximum value to the original sample. Median values are used as punctual estimations. Red, blue and orange represent the number of regions using 61, 40 and 21 gradient directions respectively.

Figure 1. Visual comparison of the results with the diverse diffusion encoding schemes. Results with the original sample comparing axial diffusivity between chronic and episodic migraine patients are shown. The first column contains the results for 61 gradient directions (37 regions with significant differences), and the second and third for 40 (27 regions) and 21 directions (20 regions) respectively. White matter skeleton is shown in blue and voxels with significant differences in red-yellow. The color bar shows the 1-p values (family-wise error corrected).

Fig 1a. TBSS results of FA, MD, AxD and RD images between normal controls and AD patients. Green represents mean FA skeleton of all participants; yellow-red, red, copper, and blue represent regions with decreased FA (1st row), increased MD (2nd row), increased AD (3rd row) and increased RD (4th row) separately in AD patients (P < 0.05, TFCE corrected for multiple comparisons)

Fig 1b. TBSS results of FA, MD, AxD and RD images between normal controls and MS patients. Green represents mean FA skeleton of all participants; yellow-red, red, copper, and blue represent regions with decreased FA (1st row), increased MD (2nd row), increased AD (3rd row) and increased RD (4th row) separately in MS patients (P < 0.05, TFCE corrected for multiple comparisons).

Table 1. Tissue characterization ability between mean value and CV by DKI. The number represents the total numbers in eight maps calculated from diffusivities and kurtosis, derived from DKI.

Figure 2. Diffusional heterogeneities of DKI were tested for the tissue characterization.

Figure 3: Change in DT measures at different time points. Blue, green, and orange colors denote values at week 0, week 1, and week 4 respectively. Statistically significant change is represented by symbol as ●: (W0~W1), ◆: (W1~W4), *: (W0~W4).

Figure 1: Representative FA, MD, AD, and RD maps from same cross-section of a mouse brain.

Figure 3 Mediation analysis for the associations among KIBRA, MK, RK, ALFF in the left precuneus and working memory.

Figure 1 Comparison of diffusion and kurtosis metrics between KIBRA C and TT groups.

Fig. 1: Comparison of the spine MR images and tractography for a SCI patient (B) and a matched healthy control (A). The red circle indicates a significant signal dropout at the C4 level, which matches the location of the implanted hardware in this SCI patient.

Fig. 3: Above Injury Level. Mean values of AD, FA, MD, and RD above the area of injury across five visits. Values represent mean ± SEM. There were significant differences in AD and FA between groups, indicating by “*” (p < 0.05).

Figure 1: Images of a 65-year-old woman with a parathyroid adenoma (green circles and white arrowheads). The mean values of IVIM-D, IVIM-F, IVIM-D*, DKI-MK, DKI-MD and ADC of the ROIs (green circles) in the lesion were 0.70×10^-3mm²/s, 16.3%, 9.81×10^-3mm²/s, 1.60, 1.62×10^-3mm²/s and 0.95×10^-3mm²/s. Signal intensity of the lesion (arrowheads) decayed with b-values going up from 0 to 2000 s/mm².

Figure 2: A 72-year-old man with malignant nodule on left thyroid lobe. a-c: fat-suppressed T2-weighted images of the lesion (arrowheads) at the position of axial, coronal and sagittal. d-f: ROIs (green circle) of the lesion shown on IVIM parametric maps (d: D map, e: F map and f: D* map). g-i: DKI parametric maps (g: MK map and h: MD map) and ADC map (i) of the lesion (asterisks). The mean values of D, F, D*, MK, MD and ADC of the malignant lesion were 0.89×10^-3mm²/s, 26.20%, 7.05×10^-3mm²/s, 0.87, 2.08×10^-3mm²/s and 1.19×10^-3mm²/s, respectively.

Fig.1. Column bar graph (Mean with SD) of the parameters (f, D*, D, Dapp, Kapp and ADC) of IVIM-DWI and sDKI models in psoas group, retroperitoneal liposarcoma (RPLS) and its subgroups (well-differentiated liposarcoma (WDLS), dedifferentiated liposarcoma (DDLS), myxoid liposarcoma (MLS)), which were analyzed by LSD test. (ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001)

A 34-year-old woman with a large retroperitoneal soft tissue mass. mDXION-water phase showed iso-signal or low signal, while T2WI/SPAIR image showed high signal. There were more mucous components in the lesion, and the solid components were obviously enhanced after enhancement, while the mucous area showed uneven flocculent enhancement. IVIM-DWI and sDKI parameter maps: mean f value=62%, mean D value=2.02 mm²/sec×10^-3, mean Kapp value=0.33. Postoperative pathology confirmed MLS.

Figure 1. Illustration of manual contouring of a parotid tumor (blue) on EP-DWI (b₀) for further calculating the mean and standard derivation on ADC map.

Figure 2. ADC of parotid gland tumors of PMA, WT and MT on the largest slice and whole tumor. PMA: pleomorphic adenoma; WT: Warthin’s tumor; *** represents a P <0.001.

A 65-year-old woman with pathologically confirmed epithelioid angiomyolipoma in the left kidney. The lesion showed low signal intensity (SI) on T2-weighted image (a). (b-g) Corresponding parametric maps(ADC, D, D*,f, MD, MK). Tumor values (white arrow) were 0.51×10⁻³ mm²/s, 1.04×10⁻³ mm²/s, 57.8×10⁻³ mm²/s, 0.24, 1.67, 0.77, respectively. (h) pathological analysis displayed changes indicating ccRCC (Fuhrman II) (hematoxylin and eosin, ×200).

A 66-year-old man with pathologically confirmed clear cell renal cell carcinoma in the right kidney. The lesion showed very low signal intensity (SI) on T2-weighted image (a). (b-g) Corresponding parametric maps (ADC, D, D*,f, MD, MK). Tumor values (white arrow) were 0.32×10⁻³ mm²/s, 0.92×10⁻³ mm²/s, 10.09×10⁻³ mm²/s, 0.43, 1.05, 1.01, respectively. (h) pathological analysis confirmed ccRCC (Fuhrman IV) (hematoxylin and eosin, ×200).

Figure 1. Qualitative differences are apparent in multi-compartment indices between controls (N=21) and MS patients (N=12). Indices are registered to template space and averaged. For DTI (top), NODDI (middle), and SMT (bottom), indices for control and MS are displayed on the same scale.

Figure 3. Multi-compartment indices derived from DTI (top), NODDI (middle), and SMT (bottom) are plotted against EDSS score for all MS patients (Note that 10 of the 12 MS patients had clinical EDSS scores). Lines of best fit are shown over the whole cord (black), and in WM (red) and GM (blue) regions. No correlations were statistically significant – although clear (nonlinear) trends are apparent in many indices.

Figure1. Representative slices of the axial, apparent diffusion coefficient (ADC) maps generated by different combinations of b-values: (a) 0 and 1000s/mm2 (ADC1), (b) 100 and 1000s/mm² (ADC2) and (c) 100 and 1500s/mm² (ADC3). For each map, the location of the cancer region is identified with an orange arrow.

Figure2. (a) A mean squared error plot as a function of the log of a LASSO regularization parameter λ for ADC2. The dotted lines illustrate each fold, and the solid line represents the average across 5 fold cross validation. The lambda with the minimum mean squared error was chosen as the final regularization parameter (vertical line). (b) LASSO coefficient profiles of the radiomics features for ADC2. A coefficient plot was produced against the log of lambda sequence. A vertical line was drawn at the log of lambda chosen in (a), where optimal λ resulted in 4 nonzero coefficients.

Figure 1. Two single-voxel data sets that resulted in a narrower (a,c) and a wider (b,d) b-range in D quantification. Starting with a linear regression of log(S/S₀) vs. b for b-values within 400-1000 s/mm², the R² (red symbols in c, d) were calculated. Each subsequent linear-regression fitting involved an additional data point from the previous DWI frame (smaller b value). a,b: demarcate “optimal range” as black symbols with linear-regression lines. c,d: show R² plot against addition b points (abscissa direction: right to left). The maximum R² values are indicated with blue arrows.

Figure 2. a: ADC map from one subject. b: b_min map determined using the R² approach. c: a binary map for pixels with AIC model selection favoring the two-step modeling approach. The lesion ROI is indicated in red.

Figure 2: Echo time dependence on $$$T_2w$$$ and $$$ADC$$$. (A) Diffusion-free $$$T_2w$$$ images $$$S|_{b=0}(T_E)$$$ and (B) $$$ADC$$$ maps for increasing $$$T_E$$$ for patient 1. For each image, the corresponding histograms are shown for benign transition zone (blue) and PIRADS-5 lesion (red). The mean and standard deviation of the $$$T_E$$$ dependence on (C) $$$T_2w$$$ and (D) $$$ADC$$$ is extrapolated for benign and malignant transition zone. At long $$$T_E$$$, the benign tissue $$$ADC$$$ increases, while the lesion $$$ADC$$$ is nearly constant.

Figure 3: Area under the curve (AUC) plotted as a function of TE, separating between benign and malignant transition zone in 3 patients.

Figure 1. The ADC value and TIC curve of breast tumor

Table 1. MRI manifestations and clinical information of ALNM (+) and ALNM (-)

Representative tractography from crush and partial cuts nerve (25%, 50% and 75%). Color-coded tracks of FA, FEFA (V1*FA), track length and MD shown.

Split violin plots of A) FA, B) RD, C) AD for sham, crush, and partial cuts nerves at 4 (left, purple) and 12 weeks (right, yellow). Boxplots for each cohort/time are given in red for comparison along with the p-values for Wilcoxon rank-sum tests between 4 and 12 weeks. Note that while significant differences were observed for all cohort/indices across the two times, this does not tell the complete story given the heterogeneous recovery observed in the partial cut samples at 12 weeks (broad distributions shown in the violin plots).

Figure 1: Figure shows example of the ADC, T2 and FA maps of a single subject.

Figure 3: Box-plots summarising results over the 9 singleton pregnancies. Each plot shows: the median (red line), the 25th and 75th percentile (grey box) and individual means of each subject ROIs (circles).

Figure 4: Dynamic changes of FA value after TSCI. ROI 1 to 4 represent injured and uninjured side of the injury site and rostral to the injury site. A. Group C: The FA value of all ROIs decreases with time, probably for limited self-repair ability to recover from SCI. B. Group T: Changes of FA value in the three uninjured areas are consistent and is higher than that in the injured area, where it increases with time. C. Group M: The FA value of the injured area increases with time, suggesting that MSCs may repair injured fiber tracts. D. The DTT modeling can observe that the fiber tracts are broken.

Figure 5: Pathological and immunohistochemical examination results. T6-28d: CD90+CD105 immunofluorescent staining confirms that there exists SPIO in living hUC-MSCs. T3-75d: CD90+CD105 immunofluorescent staining finds that MSC presents abnormal living cell morphology. It can be speculated that transplanted MSCs have died or differentiated, and this needs further confirmation. T5-105d: PB staining finds that iron exists in the spinal cord gap rather than in the living cells. C2-120D: No blue-staining areas can be observed and this is a blank control group.

Fig. 2 A rabbit with a high-fat/cholesterol diet feeding for 8 weeks. (a) axial imaging with b value of 0 s/mm2, ADC, D, D*, f, DDC and α pseudo-color maps (b-g) from monoexponential, biexponential ,and stretched-exponential DWI models were 1.002×10-3mm²/s, 0.812×10-3mm²/s, 13.731×10-3mm²/s, 29.102%, 0.763×10-3mm²/s, 0.859, respectively. Histological specimen (h) (magnification, ×200). S=stomach, L=liver, SC=spinal canal.

Fig.3 Compare receiver operating characteristic curve for (a) bi-exponential vs. mono-exponential model; (b) stretched-exponential vs. mono-exponential model in distinguishing NASH from borderline or less severity group.