Figure 1. Pipeline used in the MODEM machine learning analysis. SMOTE = synthetic minority oversampling technique.

Figure 2. A set of first-order feature maps from one representative patient. The color-coded regions indicate the cancerous and normal ROIs on the diffusion-weighted images with a b-value of 800 s/mm².

Figure 2 The AUC of R2*, T2* and amplitude is 0.851, 0.843, 0.729, respectively, with P value of 0.000, 0.000, 0.001.

Table 2 Summary of mean amplitude, phase, R2* and T2* values between CA and SCC groups

Fig.1 Treatment course and MRI acquisition

The treatment schedule and the timing of MRI acquisition are shown.MRI examinations with DWI were performed a total of 4 times during the clinical course: before the start of treatment, just before the start of IGBT, at the end of EBRT, and 2-3 months after the end of treatment. The entire tumor was set as the VOI in the axial images of the ADC map. Eighteen histogram features were extracted from each VOI.

Fig.5 Kaplan-Meier plot and the p-values

Kaplan-Meier plot and the p-values from the log-rank test on disease-free survival. Kurtosis change can divide between better and worth prognosis groups.

The T2 values of all lesions under both 1.5T and 3.0T MRI (**, p value <0.01)

The T2 values of normal tissue under both 1.5T and 3.0T MRI (**, p value <0.01)

Table3:*P < 0.05 was considered statistically significant different.

Figure3: Receiving operating characteristic (ROC) analysis of ADC(T),FA,VRA value in differention of cell-rich uterinefibroids And Uterine sarcoma.

Table 1. Comparison of diagnostic performance of the radiologist and CNN models in ovarian mass discrimination based on MR images in the testing group.

Figure 3. The segmentation results on MR images. The segmented ovarian tumor regions by U-net++ network (the red line) and radiologist (the green line, ground truth) are shown on sagittal T2WI (upper row) and T2WI coronal (down row) MR images.

Figure 1: Representative T₂-weighted images of PDXs implanted in kidney overlaid with ADC, urea AUC and k_pl maps (A). The tumor is delineated with red line. Mean apparent diffusion coefficients, urea AUCs and k_pl of kidney (B), bone (C) and liver tumors (D). (Note: significance shown as p values. * p<0.05, **p<0.01 and ***p<0.001)

Figure 3: Representative T₂-weighted images of LTL610 PDXs implanted in bone at baseline and post-carboplatin overlaid with ADC, urea AUC and k_pl maps (A). Mean apparent diffusion coefficients (B), urea AUC (C) and k_pl (D) at baseline and post-carboplatin. (Note: significance shown as p values. * p<0.05 and **p<0.01.)

Figure 4. A summary of biopsy results from each patient following the hyperpolarized ¹³C study and radiologist targeting session. Low-grade cancer involvement was found in the cores corresponding to the C13 targets in 2 out of 3 patients. All the C13-mpMRI studies and the biopsies were safe and successful without adverse events.

Figure 1. This figure illustrates the workflow of HP-¹³C MR research targeting of the prostate, based on pyruvate-to-lactate conversion (k_PL) abnormalities. The HP-¹³C MR exam and research targeting was integrated into the standard-of-care mpMRI fusion and systematic biopsies at our institution.

Figure.1: Data acquisition and processing framework. Utilizing 3D patient-specific molds and ex vivo MRI, quantitative MRI maps of ADC and T2 were registered to the digital histopathology maps of epithelium, stroma and lumen area fractions (f_epithelium, f_stroma, f_lumen) at identical resolution of 1x1 mm². Then, texture features for characterizing PCa heterogeneity were obtained from the high resolution, co-registered quantitative MRI and digital histopathology maps.

Figure.4: Texture features were calculated for 9 prostate cancer (PCa) regions from 5 patients, based on co-registered (a) quantitative MRI ADC and T2 maps and (b) digital histopathology maps of epithelium, stroma and lumen area fraction. The Gleason Score (GS) for each PCa is also reported.

Figure 5: Overview of measured SWS (c) in patients and in the LNCap murine model. Comparison of the k-MDEV-based group mean SWS (±Std) determined in vivo on clinical scanners, show that PCa’s in their native environment⁽⁴⁾ are stiffer (3.1±0.6m/s, frequency-range 60-80 Hz) than the implanted LNCap tumors in the murine-model (1.5±0.3 m/s, frequency-range of 80-130Hz).Considering the frequency dispersion of c in the ex-vivo study, a good agreement can be found with in-vivo data. See the Result section for detailed SWS values measured ex-vivo.

Figure 1: Experimental setup. To induce shear waves the anesthetized animals are positioned directly on a passive pressurized air driven actuator (size:80×40×10mm³). The acquisition coil, here sketched as a halved hollow cylinder (inner diameter 47mm), is positioned over the animal. The table height is adjusted so that the animals can be positioned in the isocenter of the magnet.

Figure 1. A demonstration of the T2WI, Syn T2WI, PD, T1, and T2 maps in a cervical cancer patient. ROI is placed on the cervical lesion on Syn T2WI.

Figure 2. ROC analysis for parameter maps

Figure 1. A 55-year-old female with squamous cell carcinoma (SCC), FIGO stage IIIC1. Largest slice ROI was delineated on synthetic T2-weighted image (A), the corresponding PD, T1 and T2 maps (B, C, D) were also shown.

LEGH-AC: Multicystic mass with solid components exhibiting sponge-like slight high intensity on T2WI, high intensity on DWI with relatively low ADC (1.54 x 10^-3mm²/s), maintained high signal intensity on computed DWI (b=1500s/mm²) suggesting diffusion restriction, and rapid & prolonged contrast-enhancement pattern on DCE-MRI.

LEGH: Multicystic mass with solid components exhibiting sponge-like slight high intensity on T2WI, high intensity on DWI with relatively high ADC (2.29 x 10^-3mm²/s), decreased signal intensity on computed DWI (b=1500s/mm²) suggesting T2 shine-through effect, and gradual contrast-enhancement pattern on DCE-MRI.

Fig. 1 The Steps of Image Acquisition and Statistical Analysis.

Fig. 5 Statistical Results of ROC Curves.

Panel A is the ROC curve of regression model 1 for the comparison of poorly vs. moderately groups. Panel B is the ROC curve of regression model 2 for the comparison of moderately vs. well groups. Panel C is the ROC curve of regression model 3 for the comparison of poorly vs. moderately&well groups. Panel D is the ROC curve of regression model 4 for the comparison of well vs. moderately&poorly groups.

Bladder mucosal invasion positive case: Bladder mucosal invasion was positive on rFOV-DWI, and confirmed by cystoscopy.

Bladder mucosal invasion false positive case: Bladder mucosal invasion was positive on rFOV-DWI, but not confirmed by cystoscopy (false positive). Probably submucosal invasion with bullous edema may mimic mucosal invasion on rFOV-DWI.

Table 1. Correlation with histological findings, diagnostic performance and data reproducibility

Fig 1. Representative case. 80 year’s female. Endometrioid carcinoma, Grade 3, T1a (<1/2) N0 M0. The ROIs are assigned in the uterine endometrial lesion.

Figure 1: A 55-year-old postmenopausal patient with vaginal bleeding and endometrial carcinoma. (A) Sag T2 SPAIR. (B) APTw and T2WI fusion map, APTw SI of 3.3%. (C-E) IVIM post-processed pseudo-color images, the values of D*, and the D and f values were 7.9×10-3mm²/s, 0.469×10-3mm²/s, and 17.85%, respectively. (F) HE staining results of disease pathology (pathological type: endometrial adenocarcinoma, magnification factor: ×40)

Figure 2: A 38-year-old premenopausal patient with vaginal bleeding and endometrial polyp. (A) Sag T2 SPAIR. (B) APTw and T2WI fusion map, APTw SI of 2.35. (C-E) IVIM post-processed pseudo-color images, the values of D*, D, and f were 7.0×10-3mm² /s, 0.988×10-3mm²/s, and 51.8%, respectively. (F) HE staining results of disease pathology (magnification factor: ×40)

Table

Table

Figure 3. A 71-year-old female with endometrial carcinoma. ROI was placed in endometrial carcinoma area and superficial myometrial area. Both A-C and E-G are the Ktrans, Kep and iAUC, respectively, D and H were DCE concentration curves of the mass and superficial myometrial layer. The calculated Ktrans, Kep and and Ve values of the mass were:0.085 min−1, 0.646 min−1, 0.131, respectively. The Ktrans, Kep and and Ve values of the superficial myometrial layer were:0.31 min−1, 0.981 min−1, 0.316, respectively.

Figure 1. The Kep distribution of benign and malignant endometrium lesions. The mean Kep values of these two group of statistical measurements are: benign: 0.627 ± 0.673min−1and malignant: 1.412 ± 1.143 min−1 (T test, p<0.01). The area under the ROC curve of Kep value is 0.82 which indicates that the Kep values have very good predictive ability for differentiation between benign and malignant endometrium lesions

Figure.1. Low-grade EA (grade 1, FIGO IB, Ki-67 = 10%) in a 49-year-old woman (arrowheads). (a) Map of DWI (b = 1000s/mm²). (b) Pseudo colored map of ADC, (c) Pseudo colored map of D, (d) Pseudo colored map of D*, (e) Pseudo colored map of f, (f) Pseudo colored map of DDC, (g) Pseudo colored map of α, (h) Pseudo colored map of MTRasym (3.5ppm).

Figure.2. Plots show individual data points, averages, and standard deviations of MTRasym (3.5ppm), ADC, D, D*, f, DDC, and α in high-grade (yellow) and low-grade (green) EA. *P < 0.05, **P < 0.01, ***P < 0.001, and ●P > 0.05.

.

.

Figure 2. A 47-year-old female with left ovarian cancer. (1a) T2WI image; (1b) R2* map; (1c) phase map; (1d) tumor was delineated around the edge of the tumor; (1e) AS software recognized quantitatively and automatically ITSS ratio by reading phase maps. ITSS recognized were covered in green.

Table 2. Area under curve (AUC), sensitivities, specificities of R2*, ITSS and the combination of the two parameters for differentiation between MOTs and OOTs groups

Figure 1: Representative Placental blood flow (PBF) and Arterial Transit Time (ATT) results, for one placenta (GA = 32 weeks) superimposed on the corresponding T2 anatomical image.

Figure 2: Mean Placental Blood Flow (PBF) (A) and Arterial Transit Time (ATT) (B) values of three GA. The first two early GA (mean of 15 weeks –orange, and 20.5 weeks- gray, are taken from⁵), while the late GA (mean of 32.5 weeks, blue) demonstrates our results.

a-c. Box plot of IVIM parameters for normal and FGR groups. a) standard diffusion coefficeint (D); b). pseudodiffusivity (D*); c) perfusion fraction (f). d) Scatter plot showing the perfusion fraction (f) in relation to gestational age in normal pregnancy (p = 0.01). The estimated regression line is shown in the scatter plot (R² = 0.173).

An example of IVIM result from a FGR pregnancy (38 years, 31 gestational weeks). a) diffusion weighted image at b-value = 0 mm²/s; b) perfusion fraction (f) map, with f value of 36.4%; c) standard diffusion coefficeint (D) map, with D value of 1.46×10⁻³mm²/s; d) pseudodiffusivity (D*) map, with D* value of 19.9×10⁻³mm²/s . The ROI was draw on diffusion weighted image (b = 0 mm²/s) and then transferred to the other three maps.

Figure 3. Averaged histograms of pixel-wise placental T2* values for week 14 and week 20 grouped by pregnancy complications. Here, ‘control’ means no reported pregnancy complications or outcomes and incudes obese and non-obese subjects.

Figure 2. Box-and-whisker plots of 14-week, 20-week, and ΔT2* values, with the T2* values of abnormal gestational outcome overlaid on top. A trend of increased T2* and decreased T2* with gestation age is observed for non-obese and obese subjects, respectively. However, there are no significant differences in the T2* values for the two groups.

Proton density fat fraction IDEAL image slice of (A) a control diet (CD) guinea pig, and (B) a Western diet (WD) guinea pig, with the liver segmented in red. Note the visibly elevated PDFF in the liver of the WD guinea pig. (C) Maternal liver volumes from T1-weighted images for CD and WD sows (p=0.043), (D) Proton density fat fraction (PDFF) for liver from IDEAL images for CD and WD sows (p=0.040). One star (*) indicates (p<0.05).

(A) Fetal and placental volumes from T1-weighted images for control diet (CD) and Western diet (WD) groups. (B) Placental efficiency, displayed as fetal volume/placental volume for CD and WD groups. Data points coded the same colour within each diet group are fetuses from the same sow.

Fig. 2: Estimated fetal volume as a function of gestational age (green points) overlaid on ultrasound fetal body growth chart⁴ (blue curves representing percentiles). Red points correspond to IUGR fetuses.

Fig. 1: Illustrative examples of the FIESTA (upper row) and TRUFI (lower row) fetal body segmentation results of two cases. Left: original image, middle: body segmentation result, right: manual ground truth.

Figure 1 a.) single frame from the phantom camera footage with measured segment b.) gif created from SLR footage of voiding bladder c.) gif created from MRI sequence of voiding d.) single frame from MRI sequence with measured segment.

Figure 2. a) Captures of the anterior view of the in vitro 3D segmented bladder volumes at various time points showing deformation over voiding b.) posterior view of the segmented bladder c.) animated asymmetric and complete deformation of the model at various time points.

Fig. 2- Sensitivity, specificity, and AUC of VI-RADS in identifying MIBC and NMIBC for each group. AUC = area under the curve; CI = confidence interval; MIBC = muscle-invasive bladder cancer; NMIBC = non–muscle-invasive bladder cancer; ROC = receiver operating characteristics; VI-RADS = Vesical Imaging Reporting and Data System.

Fig. 1-Design of the study. BC= bladder cancer; mpMRI = multiparametric magneticresonance imaging; VI-RADS = Vesical Imaging Reporting and Data System.

Figure 2. Muscle-invasive papillary urothelial carcinoma in the posterior bladder wall in a 71-year-old-man. a. Axial T2-weighted image; b. Axial DWI images; c. Corresponding diffusion-weighted image reconstruction of ADC values (ADC values are given in units of ×10-3 mm2/s). d. Volumetric ADC histogram shows a large portion of voxels with low ADC values and positive skewness of 2.993 and positive kurtosis of 11.225.

Figure 5. ROC curves of VI-RADS score combined with volumetric ADC histogram parameters

Figure 1 (A-D) A 58-year-old woman with low-grade BCa. (A) Axial T2-weighted image, (B-D) SyMRI-derived pseudo-colored maps (T1, T2, and PD, respectively) indicate that the mean T1, T2, and PD value are 27.27 msec, 26.18 msec, and 71.67 pu, respectively.

(E-H) A 31-year-old man with high-grade BCa. (E) Axial T2-weighted image, (F-H) SyMRI-derived pseudo-colored maps (T1, T2, and PD, respectively) indicate that the mean T1, T2, and PD value are 32.59 msec, 19.41 msec, and 71.52 pu, respectively.

Figure 2 ROC cure analysis of the mean T2 and PD values for differentiating low- from high-grade BCa. The area under the ROC curves for the mean T2 and PD values were 0.761, and 0.880, respectively.

The upper four tube showed the averaged signal intensity of APTw with urea of the concentration of 2 %, 4 %, 6 %, 8 % were 1.07%, 2.46%, 4.78%, 7.22%;the lower four tubes showed in water, egg white ,a healthy volunteer and a patient with bladder cancer were -0.38% ,8.25% 3.13% and 0.29%.

Figure 1. The APT weighted images of testes.

Pearson’s correlation showed a significant positive correlation between age and the mean APT SI of both testes.

Table 2. Comparison o MTRasym and R2* values between the two groups

Figure1 and Figure 2

Figure 2: Representative example of Four Quadrant mapping and associated metrics in a 60 years old patient with Gleason 4+5 cancer in the right peripheral zone (outlined). Histology image (A) and corresponding T2-weighted image (B), ADC map (C) with prostate quadrant map (D), angle (E) and distance (F) are shown. Cancer is associated with high PQ4. Vectors for cancer voxels have a lower angle (except AFMS) and small amplitude. The cancer had 2% PQ1, 45% PQ2, 6% PQ3 and 47% PQ4 signal voxels.

Figure 1: Four Quadrant mapping scheme where each voxel in the prostate is a point in the 4 quadrant plot, where y = slope of ADC with varying TE, and x = slope of T2 with varying b-value. Each voxel is associated with a distance from origin and an angle. Benign tissue typically lies in quadrant 2 (high PQ2; green). The distinctive property of aggressive cancers is that they have a higher percentage of voxels in the 4th quadrant or high PQ4 (red). Cancer vectors tend to have small amplitude and lie along the negative y axis.

Figure 3. Tumor probability maps for two patients across all pre-processing combinations: a) None, b) N4, c) PABIC, d) N4 + basic normalization, e) PABIC + basic normalization, f) N4 + PZ normalization, g) PABIC + PZ normalization, h) N4 + histogram normalization, i) PABIC + histogram normalization.

Figure 2. Normalization performance for basic, median peripheral zone (PZ), and histogram matching normalization. Percent coefficient of variation was employed as the ad hoc metric.

Figure 1 T2WI（A）, DWI（B）, fused T2 mapping and DWI (C, D) images , IVIM images as well as the derived Standard ADC, Fraction of fast ADC-mono, Slow ADC bi, and Fraction of fast ADC bi, et al maps (E-L) of a BPH patient. The placement of ROIs is as illustrated.

Table4CorrelationamongAPT values andthe perfusion parameters of IVIMin all patients with PCa or BPH

Figure 2. Four examples of randomly-generated synthetic prediction maps corresponding to a given ground truth map. Candidate regions in synthetic prediction maps were labeled in the same way as demonstrated in Figure 1, and radiomic features were extracted in the same way as they were for candidate regions of prediction maps obtained from the first-stage voxel-wise classifier.

Figure 1. (a) Sample ground truth and prediction map generated from the first-stage voxel-wise classifier (white = PCa). (b) Image dilation applied to maps, which facilitates identification of candidate regions in the prediction map (four in this example) via identification of connected voxels. (c) Labeling of candidate regions based on degree of overlap with voxels in the ground truth map. Candidate regions are labeled HG-PCa only if ≥ 50% of voxels within the region are labeled GS ≥ 4+3.

Figure 1 A 72-year-old man with PCa: T2WI image (A), DWI image (B, the tumor was located in the right central zone), APTw image of the prostate fused with DWI image (C, D, the placement of ROIs is as illustrated). The tumor showed the APTw value of 1.95 %. IVIM images as well as the derived maps (E-L). A 70-year-old man with BPH: T2WI image (a), DWI image (b), APTw image of the prostate fused with DWI image (c, Dd, the placement of ROIs is as illustrated). The prostate showed the APTw value of 1.31 %. IVIM images as well as the derived maps (e-l).

Figure 2 The ROC of all the parameters values in differentiation of PCa and BPH.

Figure 1.Workflow of the radiomics modeling

Figure 2. ROC curves in plan A&B, 6-1&6-2 were ROC curves for training and testing dataset.

Figure.3 Clinical dynamic contrast enhanced images of 4D-FB and e-THRIVE A. 77-year-old patient with a Gleason 7 prostate cancer, initial PSA of 4.7 ng/mL. Early dynamic contrast enhanced image (4D-FB) shows intense enhancement in the region of the right peripheral zone (arrow). B. 66-year-old patient with a Gleason 7 prostate cancer, initial PSA of 8.1 ng/mL. Early dynamic contrast enhanced image (e-THRIVE) shows enhancement in the region of the right peripheral zone (arrow).

Figure.1 schematic diagram for scan date acquisition and reconstruction in 4D -FB

Figure 1. Plots of blood S_br(t) (dots) obtained from iliac artery and its EMM fits (line) for (a) 30% of standard dose and (b) 70% dose of standard dose. (c) and (d) show plots of cancer and tissue S_r(t) (dots) and their fits (lines) obtained from the SI-Tofts model for (c) 30% and (d) 70% of standard dose.

Figure 2. Box-plots of cancer (red) and normal tissue (green) physiological parameters (a ,c) K^trans and (b, d) v_e extracted from SI-Tofts model for the 30% (top row) and 70% (bottom row) of standard dose DCE-MRI. The square (□) indicates mean and the asterisks (*) indicate the upper and lower limits of the data.

Figure 1:

A: T2w and coregistered ²³Na MRI illustrated as overlay with the TZ (white) and the PZ (red).

B: TSC quantification based on reference vials prior to image co-registration (method 1, vials encirecled in red).

C: Quantification based on VOIs within the iliac artery after image co-registration to the T2w image (method 2, iliac arteries encircled in red).

For all three images a slice was chosen that best depicts the corresponding ROI. Images show high intensity lines at the bottom which are artifacts that were caused by the B1-corrections.

Figure 3: Quantified ²³Na MRI from one patient with a PI-RADs 5 lesion in the TZ, image focused on the prostate region. TSC quantification based on method1 (left) and method2 (right).

Figure.1 A 58-year-old man with PCa in right peripheral zone, GS is 8. (a) T2-weighted image shows hypointense signal in the lesion, (b) Diffusion-weighted imaging indicates hyperintense signal, (c) APT pseudo-colored map shows yellow-green pseudocolor in the lesion, (d) Mean kurtosis pseudo-colored map indicates red-yellow-green pseudocolor in the lesion, (e) Mean diffusivity pseudo-colored map shows blue-green pseudocolor in the lesion, (f) pathological image

Figure.3 ROC curves of MK/MD/MTRasym(3.5ppm) values between BPH and PCa groups(a), BPH and low-risk groups (b), low-risk and intermediate-risk groups (c), intermediate-risk and high-risk groups (d), respectively.

Figure 1. ¹⁸F-PSMA PET/CT-US or PET/MRI-US fusion targeted prostate biopsy for the intraprostatic PET-positive lesions were performed (A). The boundaries of the prostate were delineated on the CT image (B, white circle). The PET-positive lesion was marked as the target for biopsy (C, pink circle). The previous delineated prostate and PET-positive lesion from PET/CT was registered to the prostate volume acquired from the 3-dimensional transrectal ultrasonography; the puncture needle (D, arrow) then reached the target biopsy area.

Figure 2. PET/CT found one PET-positive lesion in the prostate gland (A: PET, B: CT, C: fused PET/CT). PET/MRI showed short T2 signal (D: PET, E: T2WI, F: fused PET/T2WI), high DWI signal (G: PET, H: DWI, L: fused PET/DWI), and a decreased ADC value at the site of the PET-positive lesion (J: PET, K: ADC map, L: fused PET/ADC map). The subsequent pathology confirmed prostate cancer.

Size distribution of suspicious lymph nodes as detected by nanoparticle-enhanced MRI (green) and prostate-specific membrane antigen PET/CT (blue).

FIGURE 3. Anatomic distribution of identified suspicious lymph nodes as detected by nanoparticle-enhanced MRI and prostate-specific membrane antigen PET/CT.