Fig 3. Gaussian kernel density estimates (i.e., smoothed histograms) of awPLV between resting-state networks and gastric signals recorded concurrently (cyan) and on different days (coral). P-values from Wilcoxon rank tests have been adjusted for multiple comparisons using a FDR of 0.05. Subfigures for networks with significant gastric synchronization are outlined in red.

Fig 2. Aggregate spatial maps of resting-state networks (RSNs). Representative sagittal, coronal, and axial views (left-to-right) are overlaid on structural images in the Montreal Neurological Institute template space. Coordinates (in mm) for each view are given below each subfigure. (AUD: auditory, SMOT: somatosensory-motor, VIS: visual, DMN: default mode network, ATTN: attention, EXEC: executive, SAL: salience, CB: cerebellar, ven: ventral, dor: dorsal, r: right, l: left). Figure adapted from ¹.

Fig. 1 Simultaneous MRI acquisition and EGG recording helped in validating our newly proposed deep learning-based fully automatic GI MR image analysis pipeline. (A) layouts the scheme for the 30-channel electrode array attached over the skin surface on top of the stomach. (B) and (C) show examples of pre-processed EGG signals and acquired GI MR images, respectively. (D) displays the 2D U-net for segmenting stomach from MR images and the extraction scheme of peristaltic displacements perpendicular to the gastric long axis along the greater and lesser curvatures.

Fig. 2 The frequency and power of the estimated displacements along the greater and lesser curvatures were validated against the frequency and power of EGG, respectively. (A) shows the frequency distribution of EGG in gray lines and dots and of estimated displacements along the greater and lesser curvatures in black lines and boxes. We then targeted on 6.25 CPM, (B) shows the time-evolving power of the estimated peristalsis within the narrow frequency band along the greater and lesser curvatures in the upper panel and of EGG shown within the same narrow band in the lower panel.

Figure 1: Comparison of LFC assessed by MRI PDFF values among DM and Non-DM. The mean LFC in DM group was 12.1 % (8-20) vs Non-DM 6.7% (4.2-10.7).

Figure 2: Comparison of PFC assessed by MRI PDFF values among DM and Non- DM. The mean PFC in NAFLD subjects was 7.5% and in Non NAFLD was 6.5%.

Figure 2.

Example of ROI placement on PDFF maps of (a) the liver and (b) the body and tail of the pancreas.

Figure 4.

Graphs showing comparison of body composition parameters between the no remission and remission groups.

(a) VAT:SAT ratio (median ± IQR), (b) FM index (mean ± SD), (c) FFM index (mean ± SD), (d) SM index (mean ± SD).

Figure 3. High-resolution T2w imaging of esophagus (a and c) is further improved using DL reconstruction with commercially available AIR Recon DL software (GE Healthcare, Waukesha, WI)10. Shown are images of distal esophagus of a 37-year-old male healthy volunteer. Abbreviations: T2w double IR, T2 weighted double inversion recovery; DL, deep learning

Figure 2. Cardiac induced motion artifacts diminish image quality of T2w image of esophagus (esophagus is in the center of yellow circles) and may mimic lesion in esophageal wall. Cardiac gating and saturation bands (arrows) help to overcome this limitations. Abbreviations: T2w SSFSE, T2 weighted single shot fast spin echo; T2w double IR, T2 weighted double inversion recovery

Fig 4: Assessment of small bowel and colon texture both subjectively (top) and by Haralick contrast (bottom; median and IQR, P-values for Wilcoxon tests). When assessing small bowel and colon texture as well as their difference subjectively, all were distinctly altered in people with CF. Whilst we could not detect a difference in small bowel texture using the Haralick analysis, texture in the colon was lower in people with CF than in healthy controls. The difference between texture in colon and in small bowel were less distinct from each other in people with CF.

Fig 3: Small bowel and colon texture. Top, a coronal T1-weighted image was calculated into a Haralick contrast map. Example shown is a healthy control participant with typical textures in both small bowel (dashed arrow) and colon (solid arrows). Bottom, example curves (histograms) of a small bowel ROI of the Haralick contrast map. T1-weighted images were scored between 1 (normal, smooth) and 3 (mottled, see also Fig 3). The small bowel ROI of the resulting Haralick contrast map was then analysed, illustrating how the higher the score for heterogeneity, the wider the histogram tended to be

Comparison of the T1ρ relaxation time of the TA and SOL muscles in the three group. A) Comparison of the T1ρ map of the TA and SOL muscles in the three group. B) The T1ρ relaxation times of the SOL muscle in the three group. C) The T1ρ relaxation times of the TA in the three group.**P < 0.01 and ****P < 0.0001.

The linear relationship with the duration time of illness and fasting blood glucose of the TA and SOL muscles. A) The linear relationship with the duration time of illness of the TA muscles. B) The linear relationship with the duration time of illness of the SOL muscles. C) Fasting blood glucose levels in three groups. D) The linear relationship with the fasting blood glucose and the SOL muscles T1ρ relaxation time. E) The linear relationship with the fasting blood glucose and the TA muscles T1ρ relaxation time. **P < 0.01, ***P < 0.001 and ****P < 0.0001.

Figure 1. Results from one subject from the control cohort and a second subject from the T2DM cohort. The anatomic PSIF MR image of one slice is shown, as well as the corresponding ground truth segmentation and CNN segmentation results. The fat fraction map of the same slice is also shown. For each subject, a graph is depicted on the right, highlighting the variation in fat fraction as a function of slice location along the calf for the gastroc medial, gastroc lateral, and soleus muscles.

Table 1. Participant characteristics for the training (N=19) and fat fraction evaluation (N=37) cohorts. Fat fraction percentages were calculated over an 8.1cm longitudinal region using the automated segmentation tool.

Figure 3. OEF in the medial gastrocnemius muscle at the centrally acquired slice overlaid on an anatomical reference image in a representative subject before and after exercise training.

Figure 2. Mean oxygen extraction fraction and maximal exercise capacity (VO₂Peak) before and after supervised exercise training. Error bars indicate standard deviation.

Figure 1. Anatomical compartments (a-b) of the human the skin using 1H 9.4T MRI approaching histological length scales (20 mm) for control (NDB2) and diabetic (DB3) skin biopsies. Segmentation of dermis layer in in control (c) and diabetic (d) biopsies with the increase of TE to 8ms.

Figure 2. (a) - 1H high resolution axial plane MRI of control (NDB8) skin biopsy using gradient echo MRI protocol with TE = 2ms (1H GE). Enhanced segmentation of the dermis layer (appears dark in 1H SE) was obtained with spin-echo MRI using the same as in 1H GE geometry; (b) – free sodium image of control biopsy; (c) - 1H GE of the diabetic skin biopsy DB3 in the axial plane with corresponding free sodium 23Na MRI and co-registration image; (d) - coronal projection DB3 biopsy. Free and stored sodium spaces are co-registered within the dermis compartment.

Comparison of tissue FF and clinical indicators between Group 1 and Group 2

Comparison of tissue FF and clinical indicators between Group 2 and Group 4

Fig.3 Auto segmentation results are shown in a. original water image; b. auto segmentation results; c. artificial segmentation results, respectively. Loss values, accuracy rate, rr values are shown in the line chart

Fig.2 Image pre-processing framework

Figure 1

Figure 2. Variability in parts segmentation for the first 10 out of N=20 validation subjects for the first annotations (top), second annotations (middle), and our method’s predictions (bottom).

Figure 3. Illustration of the evaluation of parts segmentation using root mean squared error (RMSE) of the Euclidean distance between the annotation boundaries (black markers) and the predicted boundaries (solid colors). Illustrations for the intra-observer distance (left arrow), as well as distance between annotation 1 and prediction (right arrow), are shown.

Figure 1

Figure 2

Figure 1 Pancreatic fat fraction in high (>2.2ml/mU) HOMA-IR (n=8) and low (<2.2 mmol/mU) HOMA-IR (n=8) in the HIV- cohort.

Pancreatic fat fraction in high (>2.2ml/mU) HOMA-IR (n=6) and low (<2.2 mmol/mU) HOMA-IR (n=10) in the cohort with HIV+.

Figure 1. PDFF and R2* maps comparisons between scans with axial, coronal, and sagittal acquisitions using optimized CSE FAM sequences. The first row includes movies from real-time SGRE acquisition. The red arrow indicates the potential missing tissue in axial acquisition due to through-plane motion during free breathing. However, in both coronal and sagittal acquisitions, the liver is captured without obvious missing tissue while performing free breathing scanning (first row), which may enable whole-liver PDFF/R2* mapping while minimizing missed tissue (second/ third rows).

Figure 3. Bland-Altman plots between PDFF/R2* measurements from 2D CSE technique and FAM CSE technique with reference PDFF/R2* obtained from 3D CSE in all three acquisition directions (axial, coronal, and sagittal). Low bias in PDFF/R2*, with narrow limits of agreement are observed for both 2D CSE and FAM CSE.

Figure 4. Qualitative comparison across tested approaches on three representative cases. (a) T1-weighted water image, (b) baseline approach, (c) deep learning approach, (d) deep learning with pediatric and high iron case augmentation, (e) manual annotation. DSC against manual annotation is provided at the bottom left corner for each case of (b, c, d).

Figure 3. Quantitative comparison across tested approaches on DSC, ASSD, and 95^th HD.

Figure 2: Multi-parametric linear relationships obtained using multi-TE (first row) and multi-TE-TR (second row) STEAM-MRS at 1.5T. The line of best fit (solid red line) and 95% confidence intervals (dashed black lines) are displayed. Individual data points are color coded based on liver iron concentration (mg of Fe per g of dry tissue) measured by Ferriscan. High (R² > 0.75) and moderate correlations (R² = 0.44 – 0.69) were observed between various parameters. Similar slopes were observed across sequences, and R1_water demonstrated less of a dependence on LIC than R2 and FWHM.

Figure 1: Multi-parametric coefficient of determination (R²) correlation matrices for parameters obtained using multi-TE (first column) and multi-TE-TR (second column) STEAM-MRS at 1.5T (first row) and 3T (second row). The subscripts W and F indicate the parameters associated with the water and fat resonance peaks, respectively. The linear relationships (slopes and intercepts) associated with these correlation matrices are given in Table 1, except for the PDFF relationships. These were excluded due to inconsistent results across the multi-TE and multi-TE-TR sequences.

Figure 2: Patient results. A: There is a statistically significantly strong correlation between HISTO MRS T₂ and FS rTSE T₂ with r = 0.71, P < 0.0001. B: a statistically not significant weak negative correlation between hepatic FF and HISTO MRS T₂ with r = -0.22, P = 0.0734 and C: a statistically not significant weak positive correlation between FF and FS rTSE T₂ with r = 0.11, P = 0.3617.

Figure 1: Phantom results. A: HISTO MRS T₂ values strongly correlate with FS rTSE T₂ values. B: Systematic difference between the two acquisitions can be observed. A decrease in T₂ values can be detected with increasing FF up to 10 %, then the decrease levels out. Phantom composition, as the samples were constructed in 50ml tubes with the given FF filled up with MnCl₂ solution, can be a potential explanation.

Figure 1. Representative multi-echo UTE liver images at varied degrees of iron overload in a rabbit model. With the iron deposition degree increasing, UTE liver signal decreased gradually. With the increase of echo times, UTE liver signal decreased gradually.

Figure 2. Correlation between hepatic R2* and liver iron content (LIC) in the normal fat liver group (A) and fatty liver group(B). The correlation coefficients were 0.911 and 0.811, respectively.

Fig.4. First row: Axial T2 maps obtained from GraSE and T2-prepared TFE. The red box represents the MRS voxel. Second row: MRS and the T2-prepared TFE coronal reformatted T2 map. Mean ROI value represents the mean value within the MRS acquisition voxel. T2 values from the imaging sequences deviate by 76 % (GraSE) and 18% (T2-prepared TFE) from MRS. Artifacts due to motion are visible in the GraSE but not in the T2-prepared TFE sequence.

Fig.3. Separated T2-weighted water images obtained from the T2-prepared TFE acquisition with DIXON readout for with an effective echo time of the T2-preparation of 15ms. The fat signal is effectively suppressed, and the resulting isotropic resolution T2-weighted image allows reformatting in all 3 planes.

Figure 4. In vivo feasibility studies demonstrate diagnostic image quality for T2 estimation in the liver using the modified phase-based T2 mapping method. An ROI drawn in the liver shows a T2 estimate of 26ms. Multi-TE STEAM-MRS shows a T2 estimate of 28ms. The phase-based T2 map is generated assuming T1 = 1000ms, while the T2 estimate in the ROI is estimated using "average first, fit second" and a T1-corrected T2 estimation.

Figure 3. Phantom experiments demonstrate improved T2/R2 estimation with the T1-corrected phase-based T2 mapping method. (a) Reference R2 map generated from a single-echo SE sequence. (b) An example of T2_PB(T1), T2_fit(T1), and the uncorrected (red arrow) and corrected (purple arrow) T2 estimate are plotted for a single vial (reference T2 = 11ms). (c) Linear regressions of phase-based R2 as a function of single-echo SE R2 for the uncorrected (assuming T1 = 1000ms) and T1-corrected phase-based T2 method are shown. Slope and intercept are shown with their 95% confidence interval.

Isolation of water T1 values using radial T1 mapping in a patient with 19% and 3% PDFF.

Comparison of the mean T1 values between techniques in patients with and without cirrhosis.

Comparisons of the diagnostic ability of three optimal parameters (75th of T1_native, entropy_20min and entropy_10min) for discriminating LF ≥ F1 (a), ≥ F2 (b), ≥ F3 (c) and F4 (d), respectively. Entropy_20min showed the highest diagnostic performance, with AUC=0.908, 0.951, 0.969, and 0.914, respectively.

Examples of ROIs drawn over T1 maps of F1 liver tissue. The liver borders, major vessels, and gallbladder were manually excluded. The ROIs were delineated to include the whole liver on all slices for T1_native (a), T1_10min (b), and T1_20min (c), and a minor adjustment was adopted to update the ROI. The histogram curves of T1_native (d), T1_10min (e), and T1_20min (f) are shown.

Figure 4: Shear velocity map in the abdomen for volunteer 1. The shear velocity values are acquired with Intenso, averaged over the four inner-most slices and matched with the abdominal anatomy (e.g. liver, spleen and spine are highlighted).

Figure 5: Elasticity and viscosity maps (KPa) in the abdomen for volunteer 1, and one component of the curl of the displacement field is shown as well. The maps were acquired with Intenso, and the viscoelastic values are averaged over the four inner-most slices. The viscoelastic maps are matching with the abdominal anatomy (e.g. liver, spleen and spine are highlighted).

The scatter plots of ΔLSM and ΔPDFF per year and their relationship in the noncirrhotic (left) and cirrhotic (right) groups. In the noncirrhotic group, low change rates and a positive correlation were observed. In the cirrhotic group, high change rates and a negative correlation were observed.

(Top) The T2WI/FS, MRE elatograms, and wave image from 3 years of follow-up examinations of a 72-year-old female with a noncirrhotic liver with typical small and positive changes in LSM, her BMI increased from 31.5 to 35.0.. (Bottom) The T2WI/FS, MRE elatograms, and wave image in a 57 year-old female with a cirrhotic liver with typical large and negative changes in LSM in one year of follow-up examinations, her BMI increased from 28.7 to 31.1.

Figure 2: MRE image (a) and UTE image (b) in a 63-year-old woman with fibrosis F1. The UTE signal is 1122 and liver stiffness was 2.55 kPa. MRE image (c) and UTE image (d) in a 48-year-old man with fibrosis F3. The UTE signal is 692 and liver stiffness was 3.11 kPa.

Figure 5. Mean UTE SIR in patients for different stages (F0, F1, F2, F3 and F4).

Fig 3. a) The tissue viscosity, GI , plotted against fibrosis labels and inflammation score. b) the shear wave absorption, α, plotted against fibrosis labels and inflammation score. c) shear wave speed, cs, plotted against fibrosis labels and inflammation score.

Fig 2. a) Anatomy and MRE transducer placement. b) ROI in magnitude image. c) shear wave speed, cs. d) shear wave absorption, α. e) tissue viscosity, GI. f) tissue elasticity Gabs.

figure1

figure2

Figure 2. Using a modified MR angiography sequence with view-sharing and parallel imaging acceleration, a reduction of pineapple juice (50% volume) can be clearly and dynamically visualized in the esophagus as it is transported by peristalsis (A). Juice from the 20 mL swallow is visible in the stomach during the 10 mL swallow, which was performed after the 20 mL swallow. Renders of the bolus created by seeded region-growing image intensity thresholding are shown next to MIPs for each time frame. Peristalsis is not well visualized by a swallow of raw pineapple juice (B).

Figure 1. A plot of the measured T1 in pineapple juice at various volume-reduction factors (left) depicts the log-linear relationship between juice T1 reduction and juice volume reduction (right). The dashed line shows a one-to-one log-linear relationship for reference, and the solid line shows the empirical fit. The Pearson correlation coefficient for juice T1 reduction vs. volume reduction is 0.996 (p<0.001).

Fig.2 The arrow highlights the contents of the esophagus in a patient with gastroesophageal reflux after drinking pineapple juice.

Fig.1 Real-time MRI of the gastroesophageal junction shows His-angle of normal volunteers (a) and patients (b) with gastroesophageal reflux during normal swallowing. The his angle was 88.6° and 119.7° respectively.

Figure 3: Free water image and 3D rendering of colonic volume taken from participants 3 and 8 at 120 and 300 after the rice pudding meal. Plots show the colonic volume and water content measurements for all participants.

Figure 2: High resolution images taken of the colon in participants 3 and 8. The colon is indicated by the small white arrows.

Fig1. Compared with HC, significant increases (warm color) in the resting-state functional connectivity of the SN in DMCN patients. (p < 0.05, GRF corrected)

Fig4. Correlations between the rsFC of the right FIC in the SN and the MoCA scores of T2DM patients (r = 0.334，P = 0.007)

Fig. 1. Pointwise comparison of fractional anisotropy (FA) and mean diffusivity(MD) along the 20 WM tract among the type 2 diabetes (T2DM), and healthy controls (HC), which has a significant differences. The FA profiles are presented in mean ± standard deviation (SD). The red line stands for the T2DM , and blue for HC. The solid lines stand for the mean values and the shaded areas for SD. The horizontal axis indicates the location between the beginning and termination waypoint region of interest along the given tract. L (R), left (right) hemisphere.

Fig. 3. The receiver operating characteristic analysis of group classification using support vector machine

Figure1: Topological demonstration of the link between POC, SOC, and AOC.

Figure2: Schematic of olfactory network

Figure 1. Regions showing CBF differences between T2DM patients and healthy controls (P<0.05, GRF corrected); the color-bar represented the T-value of the two-sample T-test, and the blue represented the lower CBF in T2DM patients than in HC group.

Figure 2. Regions showing cerebral perfusion networks discriminated T2DM patients and healthy controls (P<0.05). GIS1 (red-yellow) represented a perfusion characteristic network composed of bilateral superior occipital gyrus, cuneate lobe and superior parietal gyrus. GIS2 (blue-green) represented a perfusion characteristic network composed of bilateral anterior cingulate gyrus, caudate nucleus head, shell, et.al.

Figure 2. The diffusion parameter maps from MAP-MRI, NODDI, and DKI for a T2DM patient.

Table 1. Independent student's T-test analysis among T2DM_MCI, T2DM_UCI, and healthy control groups in pairs

Figure 1 Statistical maps showing VMHC differences between IBS and HCs (P<0.05, corrected with Alphasim).Blue denotes reduced VMHC and the color bar indicates the T value from t test between groups.VMHC= voxel-mirrored homotopic connectivity; IBS= irritable bowel syndrome; HC=healthy control.

Figure 2 Correlation analyses between FA/Fiber length and IBS-SSS score in PCG. FA=fractional anisotropy;IBS-SSS=irritable bowel syndrome severity scoring system; PCG=posterior cingulate gyrus.

Figure 2: Experimental protocol (A) baseline familiarisation and strength assessment. (B) ¹H mDIXON scans to image calf and quantify whole muscle volume and % fat fraction. (C) Non-localised pulse-acquired ³¹P MRS protocol involving continuous ³¹P spectra acquisition during the resting state, transition into ischemic plantar flexion exercise and subsequent non-ischemic exercise recovery following release of blood pressure cuff.

Figure 5 : Post exercise PCr recovery kinetics (Mean ± SEM).

Fig. 1 The ROI placed in focus area. The 38-year-old male patient with severe diabetic nephropathy. A was R2* image,the R2* value was 15.7. B was FF image,the FF value was 3.5.

Fig 2.The ROC curve of R2* and FF value to diagnose severe and mild groups of diabetic nephropathy.

Table 1. Participant characteristics, diabetic markers, and DTI markers for the cross-sectional study.

Table 2. Participant characteristics, diabetic markers, and DTI markers for the longitudinal study (N=16).

Table 1. General Characteristics of the Study Population

Table 2. Two-observer measurement consistency

Figure 2: Diffusion-weighted image and apparent diffusion coefficient map obtained with echo planar imaging with Compressed SENSE (EPICS).

Table 2: Quantitative imaging parameters in C-EPI-DWI and EPICS.

Figure 4. Representative ADC and diffusion-weighted images (b = 0 and 800 s/mm²) with BH-DWI (upper row) and RT-DWI (lower row). BH-DWI, breath-hold diffusion-weighted imaging; RT-DWI, respiratory-triggered diffusion-weighted imaging; ADC, apparent diffusion coefficient.

Figure 2. Comparisons between BH- and RT-DWI for ADC in different abdominal regions. BH-DWI, breath-hold diffusion-weighted imaging; RT-DWI, respiratory-triggered diffusion-weighted imaging; ADC, apparent diffusion coefficient.

Figure 2. Methodology of this study

(Upper-left) Visual score: susceptibility artifact was visually evaluated by using4-graded-score.

(Upper-right) Signal intensity ratio: SI (pancreas)/SI (spinal cord) was evaluated. Pancreatic head, body, and tail were measured.

(Lower) ADC values were measured in the pancreatic head, body, and tail.

Figure 4. Visual score of the pancreatic body

The visual score of EPICS were significantly higher than that of SENSE in the pancreatic body (P<0.001).

A solid and pseudopapillary neoplasm of pancreas in (a) EPICS and (b) SENSE. Tumor borders are obscured, and the tumor size appears to be smaller in SENSE than in EPICS due to susceptibility artifact.

Figure 2. Wirsungocele visualized as a saccular dilatation of the distal ventral duct just proximal to the major papilla (white arrow).

Figure 1. Santorinicele visualized as a saccular dilatation of the distal dorsal duct just proximal to the minor papilla (white arrow); ventral ductal invisible indicates a complete pancreas divisum.

Table 1: Imaging parameters for MRCP imaging sequences

Figure 1 52-year-old man with history of gallbladder Carcinoma. Comparison of conventional RT-SPACE-MRCP (A) vs. modified RT-SPACE-MRCP (B) in a patient with gallbladder Carcinoma (Maximum-intensity-projection, MIP)

NT-MRCP and BH-MRCP showed better visualization of the segment 2 and 3 branch of the intrahepatic duct than BH CS-MRCP (indicated by arrows).

Compared with NT-MRCP, the rounded marginal morphology of filling defect was was correctly identified by BH-MRCP, but was depicted as straight edge by BH-CS MRCP (indicated by arrows).