R ecent highest-level evidence is mounting that there is an unequivocal benefit of MRI-targeted biopsies replacing or complementing systematic biopsies for diagnosis of prostate cancer (1-3). Realizing the integral role of MRI in diagnosis of prostate cancer, the Prostate Imaging Reporting and Data System (PI-RADS) continues to be developed (4-7). PI-RADS allows standardization of prostate MRI interpretation, which is a difficult task due to heterogeneous signal changes from benign prostatic hyperplasia, inflammation, and scarring after biopsy mimicking or hiding the appearance of prostate cancer. Even with PI-RADS, the high level of expertise required for accurate interpretation and persistent interobserver variability (8) limit consistency and availability while the demand for prostate MRI interpretation is at unprecedented levels. As convolutional neural networks approach or exceed human performance in natural image analysis (9), they promise to revolutionize computer-aided diagnosis and are increasingly evaluated in prostate MRI (10-15). U-Net is a convolutional neural network architecture optimized for image segmentation. It consists of an encoder and decoder

Stejskal and Tanner's ingenious pulsed field gradient design from 1965 has made diffusion NMR and MRI the mainstay of most studies seeking to resolve microstructural information in porous systems in general and biological systems in particular. Methods extending beyond Stejskal and Tanner's design, such as double diffusion encoding (DDE) NMR and MRI, may provide novel quantifiable metrics that are less easily inferred from conventional diffusion acquisitions. Despite the growing interest on the topic, the terminology for the pulse sequences, their parameters, and the metrics that can be derived from them remains inconsistent and disparate among groups active in DDE. Here, we present a consensus of those groups on terminology for DDE sequences and associated concepts. Furthermore, the regimes in which DDE metrics appear to provide microstructural information that cannot be achieved using more conventional counterparts (in a model-free fashion) are elucidated. We highlight in particular DDE's potential for determining microscopic diffusion anisotropy and microscopic fractional anisotropy, which offer metrics of microscopic features independent of orientation dispersion and thus provide information complementary to the standard, macroscopic, fractional anisotropy conventionally obtained by diffusion MR. Finally, we discuss future vistas and perspectives for DDE. Magn Reson Med 75:82-87, 2016. V C 2015 Wiley Periodicals, Inc.

Purpose To compare biparametric contrast-free radiomic machine learning (RML), mean apparent diffusion coefficient (ADC), and radiologist assessment for characterization of prostate lesions detected during prospective MRI interpretation. Materials and Methods This single-institution study included 316 men (mean age ± standard deviation, 64.0 years ± 7.8) with an indication for MRI-transrectal US fusion biopsy between May 2015 and September 2016 (training cohort, 183 patients; test cohort, 133 patients). Lesions identified by prospective clinical readings were manually segmented for mean ADC and radiomics analysis. Global and zone-specific random forest RML and mean ADC models for classification of clinically significant prostate cancer (Gleason grade group ≥ 2) were developed on the training set and the fixed models tested on an independent test set. Clinical readings, mean ADC, and radiomics were compared by using the McNemar test and receiver operating characteristic (ROC) analysis. Results In the test set, radiologist interpretation had a per-lesion sensitivity of 88% (53 of 60) and specificity of 50% (79 of 159). Quantitative measurement of the mean ADC (cut-off 732 mm/sec) significantly reduced false-positive (FP) lesions from 80 to 60 (specificity 62% [99 of 159]) and false-negative (FN) lesions from seven to six (sensitivity 90% [54 of 60]) (P = .048). Radiologist interpretation had a per-patient sensitivity of 89% (40 of 45) and specificity of 43% (38 of 88). Quantitative measurement of the mean ADC reduced the number of patients with FP lesions from 50 to 43 (specificity 51% [45 of 88]) and the number of patients with FN lesions from five to three (sensitivity 93% [42 of 45]) (P = .496). Comparison of the area under the ROC curve (AUC) for the mean ADC (AUC = 0.84; AUC ≤ 0.87) vs the RML (AUC = 0.88, P = .176; AUC ≤ 0.89, P ≥ .493) showed no significantly different performance. Conclusion Quantitative measurement of the mean apparent diffusion coefficient (ADC) improved differentiation of benign versus malignant prostate lesions, compared with clinical assessment. Radiomic machine learning had comparable but not better performance than mean ADC assessment. © RSNA, 2018 Online supplemental material is available for this article.

Objective To evaluate diffusion kurtosis imaging (DKI) and magnetisation transfer imaging (MTI) compared to standard MRI for prostate cancer assessment in a re-biopsy population. Methods Thirty-patients were imaged at 3 T including DKI (K app and D app) with b-values 150/450/800/1150/1500 s/mm 2 and MTI performed with and without MT saturation. Patients underwent transperineal biopsy based on prospectively defined MRI targets. Receiver-operating characteristic (ROC) analyses assessed the parameters and Wilcoxon-signed ranked test assessed relationships between metrics. Results Twenty patients had ≥ 1 core positive for cancer in a total of 26 MRI targets (Gleason 3+3 in 8, 3+4 in 12, ≥ 4+3 in 6): 13 peripheral (PZ) and 13 transition zone (TZ). The apparent diffusion coefficient (ADC) and D app were significantly lower and the K app and MT ratio (MTR) significantly higher in tumour versus benign tissue (all p ≤ 0.005); ROC values 0.767-1.000. Normal TZ had: lower ADC and D app and higher K app and MTR compared to normal PZ. MTR showed a moderate correlation to K app (r = 0.570) and D app (r =-0.537) in normal tissue but a poor correlation in tumours. No parameter separated low-grade (Gleason 3+3) from high-grade (≥ 3+4) disease for either PZ (p = 0.414-0.825) or TZ (p = 0.148-0.825). Conclusion ADC, D app , K app and MTR all distinguished benign tissue from tumour, but none reliably differentiated low-from high-grade disease. Key Points • MTR was significantly higher in PZ and TZ tumours versus normal tissue • K app was significantly lower and D app higher for PZ and TZ tumours • There was no incremental value for DKI/MTI over mono-exponential ADC parameters • No parameter could consistently differentiate low-grade (Gleason 3+3) from high-grade (≥ 3+4) disease • Divergent MTR/DKI values in TZ tumours suggests they offer different functional information

Background Studies on intravoxel incoherent motion (IVIM) imaging are carried out with different acquisition protocols. Purpose To investigate the dependence of IVIM parameters on the B0 field strength when using a bi‐ or triexponential model. Study Type Prospective. Study Population 20 healthy volunteers (age: 19–28 years). Field Strength/Sequence Volunteers were examined at two field strengths (1.5 and 3T). Diffusion‐weighted images of the abdomen were acquired at 24 b‐values ranging from 0.2 to 500 s/mm2. Assessment ROIs were manually drawn in the liver. Data were fitted with a bi‐ and a triexponential IVIM model. The resulting parameters were compared between both field strengths. Statistical Tests One‐way analysis of variance (ANOVA) and Kruskal–Wallis test were used to test the obtained IVIM parameters for a significant field strength dependency. Results At b‐values below 6 s/mm2, the triexponential model provided better agreement with the data than the biexponential model. The average tissue diffusivity was D = 1.22/1.00 μm2/msec at 1.5/3T. The average pseudodiffusion coefficients for the biexponential model were D* = 308/260 μm2/msec at 1.5/3T; and for the triexponential model D1* = 81.3/65.9 μm2/msec, D2* = 2453/2333 μm2/msec at 1.5/3T. The average perfusion fractions for the biexponential model were f = 0.286/0.303 at 1.5/3T; and for the triexponential model f1 = 0.161/0.174 and f2 = 0.152/0.159 at 1.5/3T. A significant B0 dependence was only found for the biexponential pseudodiffusion coefficient (ANOVA/KW P = 0.037/0.0453) and tissue diffusivity (ANOVA/KW: P < 0.001). Data Conclusion Our experimental results suggest that triexponential pseudodiffusion coefficients and perfusion fractions obtained at different field strengths could be compared across different studies using different B0. However, it is recommended to take the field strength into account when comparing tissue diffusivities or using the biexponential IVIM model. Considering published values for oxygenation‐dependent transversal relaxation times of blood, it is unlikely that the two blood compartments of the triexponential model represent venous and arterial blood. Level of Evidence: 1 Technical Efficacy Stage: 2 J. Magn. Reson. Imaging 2019;50:1883–1892.

Purpose To evaluate a radiomics model of Breast Imaging Reporting and Data System (BI-RADS) 4 and 5 breast lesions extracted from breast-tissue-optimized kurtosis magnetic resonance (MR) imaging for lesion characterization by using a sensitivity threshold similar to that of biopsy. Materials and Methods This institutional study included 222 women at two independent study sites (site 1: training set of 95 patients; mean age ± standard deviation, 58.6 years ± 6.6; 61 malignant and 34 benign lesions; site 2: independent test set of 127 patients; mean age, 58.2 years ± 6.8; 61 malignant and 66 benign lesions). All women presented with a finding suspicious for cancer at x-ray mammography (BI-RADS 4 or 5) and an indication for biopsy. Before biopsy, diffusion-weighted MR imaging (b values, 0-1500 sec/mm) was performed by using 1.5-T imagers from different MR imaging vendors. Lesions were segmented and voxel-based kurtosis fitting adapted to account for fat signal contamination was performed. A radiomics feature model was developed by using a random forest regressor. The fixed model was tested on an independent test set. Conventional interpretations of MR imaging were also assessed for comparison. Results The radiomics feature model reduced false-positive results from 66 to 20 (specificity 70.0% [46 of 66]) at the predefined sensitivity of greater than 98.0% [60 of 61] in the independent test set, with BI-RADS 4a and 4b lesions benefiting from the analysis (specificity 74.0%, [37 of 50]; 60.0% [nine of 15]) and BI-RADS 5 lesions showing no added benefit. The model significantly improved specificity compared with the median apparent diffusion coefficient (P < .001) and apparent kurtosis coefficient (P = .02) alone. Conventional reading of dynamic contrast material-enhanced MR imaging provided sensitivity of 91.8% (56 of 61) and a specificity of 74.2% (49 of 66). Accounting for fat signal intensity during fitting significantly improved the area under the curve of the model (P = .001). Conclusion A radiomics model based on kurtosis diffusion-weighted imaging performed by using MR imaging machines from different vendors allowed for reliable differentiation between malignant and benign breast lesions in both a training and an independent test data set. RSNA, 2018 Online supplemental material is available for this article.

While nuclear magnetic resonance diffusion experiments are widely used to resolve structures confining the diffusion process, it has been elusive whether they can exactly reveal these structures. This question is closely related to x-ray scattering and to Kac's "hear the drum" problem. Although the shape of the drum is not "hearable," we show that the confining boundary of closed pores can indeed be detected using modified Stejskal-Tanner magnetic field gradients that preserve the phase information and enable imaging of the average pore in a porous medium with a largely increased signal-to-noise ratio.

Nuclear magnetic resonance (NMR) diffusion experiments offer a unique opportunity to study boundaries restricting the diffusion process. In a recent Letter [Phys. Rev. Lett. 107, 048102 (2011)], we introduced the idea and concept that such diffusion experiments can be interpreted as NMR imaging experiments. Consequently, images of closed pores, in which the spins diffuse, can be acquired. In the work presented here, an in-depth description of the diffusion pore imaging technique is provided. Image artifacts due to gradient profiles of finite duration, field inhomogeneities, and surface relaxation are considered. Gradients of finite duration lead to image blurring and edge enhancement artifacts. Field inhomogeneities have benign effects on diffusion pore images, and surface relaxation can lead to a shrinkage and shift of the pore image. The relation between boundary structure and the imaginary part of the diffusion weighted signal is analyzed, and it is shown that information on pore coherence can be obtained without the need to measure the phase of the diffusion weighted signal. Moreover, it is shown that quite arbitrary gradient profiles can be used for diffusion pore imaging. The matrices required for numerical calculations are stated and provided as supplemental material.