Figure 2. Quantitative maps from a 47-yo M with muscle edema pattern observed in the left infraspinatus, and anterior and middle deltoid. The MD, RD and AD maps (in µm2/ms) do not provide contrast in denervated muscles (red arrows) vs. non-denervated muscles (green arrows), but FA, AFD (µm) and T₂-maps (ms) do. No fatty infiltration was observed in the FF (%) maps. Most pixels in the AFF (%) were mostly low (<20%). The diffusion maps were overlaid on diffusion trace images, while the T₂-map was overlaid on the T₂ magnitude image, and the FF map was overlaid on the Dixon water image.

Table 2. Summary of in vivo results comparing healthy controls vs. non-denervated patients (Non-den.) vs. denervated patient muscles (Den.), with mean and p-values from two-sample t-tests (α=0.05), and statistical significance with Holm-Bonferroni correction for multiple comparison indicated (by *).

Figure 2: a) Image series from numerical leg #1 using 8 spokes per frame. b) Computed (black circles) and theoretical (red lines) real and imaginary parts of the signal evolution measured in a muscle ROI. c) FF and T1_H2O maps, and difference maps (versus the theoretical maps) obtained without correction of the aliasing artifacts. d) Computed (black circles) and theoretical (red lines) real and imaginary parts of the corrected signal evolution measured in the same ROI. e) FF and T1_H2O maps, and difference maps obtained on leg #1 with correction of the aliasing artifacts.

Figure 3: a) Influence of the number of spokes used for reconstructing the MRF T1-FF image series on the Pearson correlation coefficient (r²), slope and intercept of the linear correlation, bias and 95% confidence interval (CI) between the computed and the theoretical FF and T1_H2O values. b) Influence of the number of SVD components used for dictionary matching on the Pearson correlation coefficient (r²), slope and intercept of the linear correlation, bias and 95% confidence interval (CI) between the computed and the theoretical FF and T1_H2O values.

Fig. 2) Sagittal (top) and axial (bottom) phase contrast velocity images (left) and local expansion (middle) and contraction (right) rate maps of a representative subject for the load condition during maximal dorsiflexion velocity.

Fig. 4) Line graphs showing local contraction and expansion rate during dorsi-/plantarflexion without and with load for five muscle segments. Individual subjects are indicated with grey lines (connecting with and without load conditions) and in black the mean and standard deviation of the group per segment and per load condition. Significant differences in local expansion and contraction rates between proximodistal segments are shown with hooked lines. Additionally, at the bottom, muscle area in number of voxels is shown as a percentage of muscle length (distal 0-100 proximal).

Figure 2: Single-voxel downfield spectra. A) Position of the voxel from which spectra are collected. Representative downfield spectra under selective (B) and non-selective inversion (C) conditions show potential ATP resonances at 8.2 and 8.5 ppm, and a carnosine resonance at 7.9-8.0 ppm.

Figure 4: Apparent T1 of downfield resonances after selective and nonselective inversion. For each resonance, the T1 under nonselective inversion was significantly longer than under selective inversion. * indicates p<0.01.

Figure 2. Scans from a 30-year-old female healthy volunteer. Axial T1-weighted image (A) and axial STIR (B). VL = Vastus lateralis; RF = Rectus femoris; VI = vastus intermedius; VM = Vastus medialis; AD = adductor magnus; GA = gracilis; BF = biceps femoris; ST = semi-tendinosus. Axial MBVa map (C) shows the distribution of arteriolar blood volume in normal muscle. The color scale shows the red end of the scale indicating an increasing MBVa and the blue end a decreasing MBVa.

Figure 1. Arteriolar blood volume in normal muscles (N=176), morphologically-normal appearing muscles (MN muscles) (N=67), edematous muscles (N=86) and atrophic or fat-infiltrated muscles (AF muscles) (N=47). MBVa_max represents maximum arteriolar muscular blood volume, MBVa_mean represents mean arteriolar muscular blood volume. *p < 0.05 for MN muscles, edematous muscles and AF muscles compared to normal muscles. ●p < 0.001 for AF muscles compared to MN muscles. ■p < 0.001 for AF muscles compared to edematous muscles.

Figure 5: A: Axial diffusivity maps for LLS, NLS and the model-based preprocessing approach and for different amounts of SMAM-affected DTI (from 10% to 100%). B: 2D probability density function for the location of the SMAM occurrences in the right lower leg. The high activity rate is reflected by large deviations in the AD maps in the region of m. soleus and m. gastrocnemius medialis.

Figure 1: A: Concept for SMAM generation and insertion into an undistorted DTI data set. SMAM activity rate and location can be estimated from DWI and transferred on the DTI data sets. B: Training of the mask generator with a skeletonized label mask m and Gaussian noise. C: Training of the generator for SMAM insertion with a separated background and foreground loss to preserve background signal intensity.

Illustration of pipeline and steps of mask alignment of longitudinal data

The methodology of alignment of imaging stacks is illustrated by the steps involving conversion of masks; initial and final masks in panel A and automated steps in mask alignment in panel B. In panel C, the images of time-point A and time-point B are rendered red and blue, to illustrate the incongruency between datasets before (left) and after (right) rigid, affine and b-spline registration. The non-corresponding regions can be identified as they maintain their respective color.

Histogram and plots of qMRI parameters of treatment-naïve cohort at both time-points

In the upper row, MD, FA and T2 are plotted against fat fraction, with each of the individual datapoints as dots, reduced to an average line using local regression with 95%-CI (shaded area). The bottom row represents the histograms of each of the qMRI parameters, the red arrow indicates significant changes and its direction. Time-point A is indicated in grey, and time-point B in green. All presented data is from the treatment-naïve cohort.

Figure 2. T1ρ maps of skeletal muscle in a control (a, d), a patient with CKD (b, e), and a patient with dialysis (c, f). The first row shows the thigh (a-c) and the second row shows calf (d-f). CKD: chronic kidney disease.

Figure 4. T1ρ, MRS and DTI assessment of skeletal muscle function in controls, patients with CKD and dialysis. (a). T1ρ. (b). The ratio of EMCL/IMCL. (c). Apparent diffusion coefficient (ADC). (d). Fractional anisotropy (FA). Correlation of ADC (e) and FA (f) with T1ρ in the Thigh of all subjects. CKD: chronic kidney disease; Gastro: gastrocnemius. EMCL/IMCL: extramyocellular/intramyocellular lipids. * and ** indicate significant difference with 0.01 < p < 0.05 and p < 0.01, respectively, using Mann-Whitney U test.

Figure 5: I_singleMU A) Video examples of a motor unit (MU; delineated in blue) in a DW image (top) and PC image (bottom). B) The signal profiles of the MUs in A. Top: Normalised DW latency profile (top), velocity profile (middle) and displacement profile (bottom) displayed against time post-stimulus. Start and end of contraction are depicted in circles. C) The spatial distribution of DW and phase changes for two MUs. Right image: Two MUs are seen on the DW image (white arrows) and a third MU appears visible on the phase image (grey arrowhead).

Figure 1: A) Three theoretical types of muscle contraction. The light orange bar represents the muscle fibre (or set of elements of magnetisation) before contraction and the dark red bar the result after contraction. B) Overview of the computational model, including model input and output (top), and the outcome measures from the simulated net magnetisation and cumulative phase (bottom). PGSE-DWI is the PGSE diffusion weighted imaging gradient and Bipolar PC is the bipolar phase contrast gradient.

Figure 1: Histopathological features and the corresponding qMRI Map

Figure 2: Overview of the quantitative maps of all patients (axial slice of the leg muscles)

Figure 2. Maximum strain values for FSHD patients (blue-left) and healthy controls (red-right) for 2 different time points.

Figure 3. Maximum strain values for FHSD patients that were scanned at least three times. At time points 1-4, the principal strain value was respectively t0 (mean=0.119), t1(median=0.058), t2(median=0.092), t3(median=0.073).

Figure 1. General workflow of data analysis. Mask difference between diffusion-weighted images and water images are shown. Eddy current improves the morphological match between diffusion-weighted and water images.

Figure 2. Example of comparison of derived indices from muscle diffusion tensor with and without eddy current correction in tibialis anterior (TA) muscle. (A) Comparison of FA map, MD map, color-encoded map of the first eigenvector (ε₁), and projection of ε₁ on unit hemisphere within the superficial and deep compartments of TA. Color scheme of direction encoding: red: x-direction, green: y-direction, blue: z-direction; superficial and deep compartments of the TA muscle are delineated by yellow and aqua boundaries in ε₁ maps, respectively. (B) Comparison of map of eigenvalues.

Figure 1. (A) Coronal image of mouse hindlimb with a 1cm long landmark to display the length and location of ³¹P- saddle coil; (B) 3D ³¹P-MR fingerprints and 2D ³¹P-MR fingerprints matched to a dictionary entry (C) 3D and 2D dynamic ³¹P-MR spectra

Figure 3. (A) PCr recovery rate between round 1 and round 2 stimulations (red color indicates increase in PCr recovery rate and black color indicates decrease in PCr recovery rate); (B) PCr level at the end of stimulation-induced muscle contraction and at after 16 min recovery.

Figure 1: Representative tissue structure that will be used to characterize RCI in silico, as a function of ALS progression. Top row shows a 2D section of the 3D structure below.

Figure 2: Example concentration time curves (A), ΔR₂^* time curves (B) and the corresponding TRATE values (C) over four diseases progression time points (TP) along with a normal myofiber case. Note each profile in ALS are dissimilar in shape and magnitude both for ΔR₂^* and concentration.

Figure 2: Individual results of MD from each participant by time. Orange points are pre-exercise (plotted from time of first scan), blue points are post-exercise (plotted from time exercise ceased). Error bars are one standard deviation shown in one direction only for clarity.

Figure 3: Individual results of FA from each participant by time. Orange points are pre-exercise (plotted from time of first scan), blue points are post-exercise (plotted from time exercise ceased). Error bars are one standard deviation shown in one direction only for clarity.

Table 1-4 and Figure 1

Figure 5. (A,B) pH values (mean, std) derived from the co-localized (orange) ¹H and (blue) ³¹P spectra averaged over 1-min intervals during the exercise paradigm for the gastrocnemius (left column) and soleus (right column) muscles. The 13-min exercise period is shown (shaded area).
(C,D) Linear correlations and (E,F) Bland-Altman plots between pH_1H and pH_31P values measured during the exercise paradigm of all subjects. LoA = Limits of agreement.

Figure 1. A. Pulse sequence chronogram showing the order of acquisition for each voxel and nucleus. The RF power was calibrated for each voxel prior to the dynamic acquisition. B. The volumes-of-interest, VOI 1 and VOI 2, were placed in the gastrocnemius and soleus, respectively. Typical voxels dimensions were 2x6x7 cm³. The nominal regions affected by the excitation pulses are shaded.

Figure 4: Segmentation results of three muscle groups (quadriceps: yellow, hamstrings: green, others: red).

Table 1: Dice Similarity Coefficients

Fig. 1. Representative lower-leg Dixon images and STE-DTI parameter maps from a 59-year-old Becker muscular dystrophy (BMD) patient (a) and a 58-year-old healthy control (b). STE-DTI data were acquired with a diffusion time, Δ, of 330 ms, and mean, axial, and radial diffusivities are shown in units of μm²/s. The BMD patient shows severe fat replacement in the gastrocnemius medialis and lateralis, and peroneus longus and extensor digitorum longus muscles (fat fraction = 83, 77, 77, and 64%, respectively). This leads to signal voids in the comprehensively-fat-suppressed DTI data.

Fig. 4. Fibre diameter, a, and membrane permeability, κ, from the random permeable barrier model in Becker muscular dystrophy patients and controls. Boxplots (top) show a and κ over the soleus (SOL), medial gastroc. (GAM), lateral gastroc. (GAL), tibialis anterior (TAN), tibialis posterior (TPO), peroneus longus (PER), and extensor digitorum longus (EDL). Median values = thick lines, hinges = 25th and 75th percentiles, and dots = raw data. Maps (middle) show a and κ in a 59-year-old patient and 58-year-old control. Density plots (bottom) show a and κ distributions from maps above.

Figure 1: Water T2 (left), Fat Fraction (middle), and relative B1 (right) maps of a patient (top) and healthy volunteer (bottom), reconstructed from a 17-echo acquisition without fat fraction constraint.

Figure 2: Water T2 maps for a patient, reconstructed from different numbers of echoes with (right) and without (left) an external fat fraction constraint.

Figure 1. An overview of the individual steps used for strain quantification in a representative dataset. The original magnitude image (A), the Fourier Transform (FT) of the magnitude image (k-space) (B),an elliptical filter used to isolate the first harmonic peak (C), the inverse FT of the modified k-space (D), the unwrapped phase image (E) and the quantitative strain map (F).

Figure 3. The actual and measured strain for the positive and negative simulated strain levels (Strain 1= black; Strain 2 = gray; Strain 3 = blue; Strain 4 = red) for each of the dynamics using the optimal elliptical filter size. The actual strain values are shown with dotted lines in the graph (Actual strain 1 (+/- 0.031) = black; Actual strain 2 (+/- 0.073) = gray; Actual strain 3 (+/- 0.115) = blue; Actual strain 4 (+/- 0.156) = red) . Each dot is the mean over the participants. Deformation in X-plane is shown on the left, in the Y-plane in the middle and in the Z-plane on the right.

Figure 2: Cine phase contrast MRI experiment with voluntary motion in a 11 y girl with diparetic cerebral palsy. Top left: visual paradigm for instruction and feedback for the patient. Top right: time courses of the force on the pedal with the mean force curve (over the whole experiment) in black. Bottom left: velocity vectors in the ROI in the calf at release. Bottom right: Time course of the velocity magnitude (ROI median) showing several peaks.

Figure 1: Cine phase contrast MRI experiment with synchronized electrical muscle stimulation (EMS) in a 11 y girl with diparetic cerebral palsy. Top left: current amplitude of the EMS cycle. Top right: time courses of the evoked force on the pedal with the mean force curve (over the whole experiment) in black. Bottom left: velocity vectors in the ROI in the calf at release. Bottom right: Time course of the velocity magnitude (ROI median) showing two peaks at contraction (left) and release (right).

Figure 1: Example Dixon and IDEAL-CPMG T2-water maps in a healthy volunteer (HV), a person with CMT (CMT1A) and a person with IBM (IBM). Segmentation ROIs are shown in the centre. Tibialis Anterior – red, Tibialis Posterior – dark pink, Medial Gastrocnemius – yellow, Peroneus Longus – green, Soleus – blue, Lateral Gastrocnemius – purple. Red arrows indicate areas of raised T₂ corresponding to oedema. Areas of high fat such as the bone marrow yield noise in the T₂-water maps as there is insufficient water in this area for accurate estimation.

Figure 2: Box plots of T₂-Water in individual muscle ROIs by disease type. The central mark indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. Whiskers are min and max. T₂-water values show a narrower range in comparison to Dixon-FF% in each disease, especially in the medial and lateral Gastrocnemius, Dixon-FF% has a wider spread of fat fractions for both diseases. T₂-water gives a more definitive separation of disease type and may show a higher specificity in a larger cohort.

Recovery time course for a single voxel in the lateral gastrocnemius. Without denoising (left), denoising method 1 (center), denoising method 2 (right).

MTR_asym maps without denoising (top row), with denoising method 1 (middle), and with denoising method 2 (bottom)

Figure 2. A 36 years old healthy volunteer 2D CSI voxel location and selected single voxel spectrum and zoomed Pi region from the tibialis anterior (red), soleus (yellow) and gastrocnemius (blue) muscles

Figure 1. A 36 years old healthy volunteer 2D CSI voxel location and multivoxel spectra on top of T1W MRI

Figure 4: A) MU difference maps overlaid onto the MUMRI images for each of the extracted MUs. Two from the forearm in the flexor carpi ulnaris muscle and one from the hand in the abductor pollicis muscle. B) Bar charts showing MU metrics: cross sectional area (CSA) shown in green, maximum Feret diameter shown in red and minimum Feret diameter shown in blue.

Figure 2: Data processing steps to extract the single MUs. A) MUMRI image B) Images grouped into two data sets: where motor unit is inactive and active (indicated by red arrow). Graph shows single voxel time-series from the MU territory, example alternations where MU inactive (red circle), MU active (green circles). C) MU difference map with no threshold applied. D) MU difference map with 0.5 threshold, showing good agreement with initial size of MU territory in (B).

The overall workflow of fat-fraction calculation that utilizes manual muscle segmentation, shading/intensity-correction across FOV, and classification of different tissue types based on a 3-class fuzzy c-means (FCM) algorithm.

Plots of group-averaged fat-fraction and texture parameter values with similar vertical scales in order to better visualize their trends as a function of muscle grades.

Figure 2. Individual values for the normalized concentrations of IMCL_CH2 (left) and Cr (right) for the two groups together with the lower, median and upper quartiles.

Figure 1. An example of the raw spectrum together with the fitted IMCL_CH2 and Cr components. The large peak at 1.5ppm is the methylene extracellular lipid (EMCL_CH2)

Figure 2 : Quantification of muscles deformations pipeline; A : 2D+t volume of interest in sagittal view with an arbitrary selected axial slice.

Figure 3 : Typical magnitude of displacement maps. These maps were obtained for the subject closest to the mean values of the cohort; A : 3d general view of displacement magnitude within the abdominal wall muscles from end-exhalation to end-inhalation with three anatomical MRI slices; B : 2D axial views of the magnitude of displacement computed on the three previous anatomical MRI slices; C : 3D visualization of the magnitudes within each individual muscle group.

Figure 4) The mask delineates the supraspinatus (A-white background on muscle) where tractography is performed (A-colors) overlaid by a T1 VIBE image. The line separation (A-blue line) corresponds to the separation between the posterior and anterior bundle (B-arrows) where myocytes insert posteriorly (red tracks) and anteriorly (green track).

Figure 2) Bland-Altman plots of FA across three scans (pairwise comparison of scans 1-2, 1-3, 2-3) for each volunteer (1 to 6) in the supraspinatus (sup) or infraspinatus (inf). Limit of agreement (loa) is 0.071 with a bias at 0.0.

Figure 1. Scan protocol

Figure 2. A 67-year-old woman with severe back pain.

Standard sagittal STIR image (scan time, 3:30 minutes) shows diffuse hyperintensity and a low-intensity line in the 12th thoracic vertebra, suggesting an acute compression fracture. The ultrafast STIR image (0:32 seconds) shows the equivalent lesion almost as clearly.

Representative “bright-bone” ZTE MRI images of the skull and C-spine. Upper row: reformatted axial, sagittal and coronal images. Lower row: volume-rendered images.

ROC curves of the clinical and imaging predictors in diagnosing the LvOPI.

The criteria and performance of all predictors.

Fig.1MR images of the lumbar spine .Every IVD was cut into 5 uniform parts in each UTE,R2*and T1rho.

Fig.2 Post hoc multiple comparisons among three Disc degeneration groups

Fig 1. Sample segmentations of cervical (A), thoracic (B), and lumbar (C) centrum ossification center volumes.

Fig 2. Regression lines for the volumes of the cervical (A), thoracic (B), and lumbar (C) centrum ossification centers.

Fig. 1. Flowchart of the enrolled patients. GCTB, giant cell tumor of bone; MRI, magnetic resonance imaging; TES, total en bloc spondylectomy.

Fig.2. Top panel: A 35-year-old man,(A) and (B): MR images showed a mass on the T12 vertebra, bilateral pedicle and lamina with extension into the spinal canal, managed with en bloc resection (C), at a 36-month follow-up review, there was no evidence of recurrence (D), and now the patient is still on visit.

Bottom panel: A 35-year-old woman, maximum diameter of lesion is 55mm (E),with pathologic fracture of the T12 vertebra (F) , managed with en bloc resection (G). The sagittal T2-WI MR image at 12-month follow-up, recurrence was detected (H), and confirmed by pathology with puncture.

Figure.1 Male, 24 years old, BMI= 21.47kg /m2, Fig. 4A and Fig. 4B are 3D mDixon Quant water image and FF image, respectively

Table 2 The difference of lumbar vertebrae fat fraction(FF) and image quality in different AF

Figure 1 ROIs of the bilateral multifidus (MF) and erector spinae(ES) respectively were drawn on T2mapping imaging

Figure 1: UTE-MT imaging in lumbar of a 67‐year‐old male volunteer. Figure A: UTE-MT-ON; Figure B: UTE-MT-OFF; Figure C: UTE-MTR map

Figure 2: Correlation between cartilaginous endplate UTE-MTR and Pfirrmann grade of disc in 62 subjects. Cartilaginous endplate UTE-MTR showed a positive correlation with Pfirrmann grade of disc.

Figure.1 Two figures show BOLD pseudo-color images of CLBP group(A) and control group(B).Corresponding R2*values in CLBP group are higher than that in control group.

Figure.5 Mean Differences Between the Control and the CLBP Groups for FF value (%)

Figure 4. CT (A) and UTE (B) lumbar spinal bone images from 72-year-old male patient with spinal fractures. Bone fractures can be seen clearly in both CT and UTE images.

Figure 1. 3D ZTE (A) and UTE (B) sequences. The ZTE sequence utilizes a non-selective rectangular RF pulse with short duration for excitation, followed by 3D center-out radial sampling. To minimize echo time, readout gradients are turned on prior to signal excitation so that the gradient encoding can begin simultaneously with signal excitation. The 3D UTE sequence enables slab selection by using a soft-half pulse for excitation together with a slice-selective gradient. After excitation, the spatial encoding gradient is turned on and simultaneous data acquisition begins.

Figure.3 Color-coded T2-calculated, R2* and fat fraction maps. In T2-calculated maps, blue represents areas of short T2 values. T2 maps of normal (A), DLBP(B) and swimming group (C). In R2* maps, blue represents areas of lower R2* value. R2* maps of normal (D), DLBP group(E) and swimming group (F). In fat fraction maps, white represents areas of higher fat fraction. Fat fraction maps of normal (G), DLBP(H) and swimming group(I).

Figure.1 Behavioral test results of three groups within one month after operation. (A)Gait test; (B) hot-plate test; (C)acetone test; (D)tail suspension test; (E)grip strength test. Data are reported as mean ± standard deviation of mean; *p<0.05, **P<0.01(Kruskal-Wallis test).

One case in the non-PD group. (A): Axial T1WI-enhanced MRI showed a huge soft tissue mass around the vertebra before treatment; (B): Three months after treatment, axial T1WI enhanced MRI showed the lesion soft tissue mass was significantly reduced; (C): Pretreatment Ktrans was 0.342 min-1; (D): Post-treatment Ktrans was 0.121 min-1, decreased by 64.6%; (E): Pretreatment Kep was 2.445 min-1; (F): Post-treatment Kep was 0.959 min-1, decreased by 60.8%.

One case in PD group. (A): Enhanced T1WI scan axial view showed a soft tissue mass invading the spinal canal on the left side of the vertebra, and the spinal cord was compressed; (B): Axial T1WI enhanced image showed that the soft tissue mass of the lesion was slightly larger than before, and the low signal cystic area is seen behind the spinal canal; (C): Pretreatment Ktrans was 0.503 min-1; (D): Post-treatment Ktrans was 0.750 min-1, increased by 49.1%; (E): Pretreatment Kep was 2.487 min-1; (F): Post-treatment Kep was 4.187 min-1, increased by 68.4%.

Figure 1. T2 histogram-derived parameters obtained from MAGiC and conventional T2 mapping have moderate to high correlations.

Figure 2. Receiver operating characteristic curves analysis.

Figure 2. Representative bi-component analysis of UTE on L4 with four echoes.

Figure 3 ROC curve of lumbar bone marrow IDEAL-IQ for differentiating patients with 30 < eGFR < 90 from patients with eGFR < 15. The area under the curve [AUC] of L4-water-kurt and L3-R2*-max were 0.856 (p= 0.009) and 0.822 (p= 0.018), respectively. Abbreviations: L4-water-kurt, kurtosis of water image in L4; L3-R2*-max, maximum R2* value in L3

Figure 1. The deep learning framework Tensorflow was used to construct the CNN architecture. In the present work, we used the Xception architectural model, which had been already trained using images with ImageNet.

Fig2. Subtracting a 3D ultrashort Echo Time (UTE) image acquired at short TE of 0.03 ms (a) by the one acquired at long TE of 6 ms (b), a resultant subtracted 3D UTE image (c) of a lumbar spine with relatively normal cartilage endplate (CEP) is presented at sagittal view. Well structural CEP (denoted by white arrow) is hyperintense as well as adjacent hypointense vertebral endplate. The effect of nucleus pulposus, annulus fibrosus was removed after image subtraction.

Fig3. On the subtracted 3D UTE images between short and long TEs, cartilage endplates (CEPs) were classified into six grades. Representative CEP images at each grade are shown (a: type I, b: type II, c: type III, d: type IV, e: type V, f: type VI ).

Figure 2. ROC analysis of T2 and T2* histogram parameters for identifying normal (Pfirrmann grades Ⅰ and Ⅱ) and abnormal (grades Ⅲ to Ⅴ) discs.

Figure 1. Degeneration incidence of lumbar facet joint (LFJ) and intervertebral disc (IVD) in different age groups. * P < 0.05, **P < 0.01, ***P < 0.001.

Figure 1 24 gestational weeks. Figure 1A, measuring of height and length of L5 vertebral ossification body center, and height of L3-4 IVP. Figure 1B, the location of conus medullaris.

Figure 2 24 gestational weeks, the mid-sagittal plane of lumbar spine showed linear T2 hyperintensity in L2 and L4 vertebral ossification body center.

Table 1：Diffusion parameters in active group and chronic group

Table 2. AUCs for IVIM-D，IVIM-f, DKI-MD, DKI-MK and ADC