Deep Learning k-Space-to-Image Reconstruction Facilitates High Spatial Resolution and Scan Time Reduction in Diffusion-Weighted Imaging Breast MRI.

BACKGROUND: For time-consuming diffusion-weighted imaging (DWI) of the breast, deep learning-based imaging acceleration appears particularly promising. PURPOSE: To investigate a combined k-space-to-image reconstruction approach for scan time reduction and improved spatial resolution in breast DWI. STUDY TYPE: Retrospective. POPULATION: 133 women (age 49.7 ± 12.1 years) underwent multiparametric breast MRI. FIELD STRENGTH/SEQUENCE: 3.0T/T2 turbo spin echo, T1 3D gradient echo, DWI (800 and 1600 sec/mm2 ). ASSESSMENT: DWI data were retrospectively processed using deep learning-based k-space-to-image reconstruction (DL-DWI) and an additional super-resolution algorithm (SRDL-DWI). In addition to signal-to-noise ratio and apparent diffusion coefficient (ADC) comparisons among standard, DL- and SRDL-DWI, a range of quantitative similarity (e.g., structural similarity index [SSIM]) and error metrics (e.g., normalized root mean square error [NRMSE], symmetric mean absolute percent error [SMAPE], log accuracy error [LOGAC]) was calculated to analyze structural variations. Subjective image evaluation was performed independently by three radiologists on a seven-point rating scale. STATISTICAL TESTS: Friedman's rank-based analysis of variance with Bonferroni-corrected pairwise post-hoc tests. P < 0.05 was considered significant. RESULTS: Both DL- and SRDL-DWI allowed for a 39% reduction in simulated scan time over standard DWI (5 vs. 3 minutes). The highest image quality ratings were assigned to SRDL-DWI with good interreader agreement (ICC 0.834; 95% confidence interval 0.818-0.848). Irrespective of b-value, both standard and DL-DWI produced superior SNR compared to SRDL-DWI. ADC values were slightly higher in SRDL-DWI (+0.5%) and DL-DWI (+3.4%) than in standard DWI. Structural similarity was excellent between DL-/SRDL-DWI and standard DWI for either b value (SSIM ≥ 0.86). Calculation of error metrics (NRMSE ≤ 0.05, SMAPE ≤ 0.02, and LOGAC ≤ 0.04) supported the assumption of low voxel-wise error. DATA CONCLUSION: Deep learning-based k-space-to-image reconstruction reduces simulated scan time of breast DWI by 39% without influencing structural similarity. Additionally, super-resolution interpolation allows for substantial improvement of subjective image quality. EVIDENCE LEVEL: 4 TECHNICAL EFFICACY: Stage 1.

Differentiating Benign and Malignant Breast Lesions in Diffusion Kurtosis MRI: Does the Averaging Procedure Matter?

BACKGROUND: Diffusion kurtosis imaging (DKI) is used to differentiate between benign and malignant breast lesions. DKI fits are performed either on voxel-by-voxel basis or using volume-averaged signal. PURPOSE: Investigate and compare DKI parameters' diagnostic performance using voxel-by-voxel and volume-averaged signal fit approach. STUDY TYPE: Retrospective. STUDY POPULATION: A total of 104 patients, aged 24.1-86.4 years. FIELD STRENGTH/SEQUENCE: A 3 T Spin-echo planar diffusion-weighted sequence with b-values: 50 s/mm2 , 750 s/mm2 , and 1500 s/mm2 . Dynamic contrast enhanced (DCE) sequence. ASSESSMENT: Lesions were manually segmented by M.P. under supervision of S.O. (2 and 5 years of experience in breast MRI). DKI fits were performed on voxel-by-voxel basis and with volume-averaged signal. Diagnostic performance of DKI parameters D K (kurtosis corrected diffusion coefficient) and kurtosis K was compared between both approaches. STATISTICAL TESTS: Receiver operating characteristics analysis and area under the curve (AUC) values were computed. Wilcoxon rank sum and Students t-test tested DKI parameters for significant (P <0.05) difference between benign and malignant lesions. DeLong test was used to test the DKI parameter performance for significant fit approach dependency. Correlation between parameters of the two approaches was determined by Pearson correlation coefficient. RESULTS: DKI parameters were significantly different between benign and malignant lesions for both fit approaches. Median benign vs. malignant values for voxel-by-voxel and volume-averaged approach were 2.00 vs. 1.28 ( D K in µm2 /msec), 2.03 vs. 1.26 ( D K in µm2 /msec), 0.54 vs. 0.90 ( K ), 0.55 vs. 0.99 ( K ). AUC for voxel-by-voxel and volume-averaged fit were 0.9494 and 0.9508 ( D K ); 0.9175 and 0.9298 ( K ). For both, AUC did not differ significantly (P = 0.20). Correlation of values between the two approaches was very high (r = 0.99 for D K and r = 0.97 for K ). DATA CONCLUSION: Voxel-by-voxel and volume-averaged signal fit approach are equally well suited for differentiating between benign and malignant breast lesions in DKI. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 3.

Value of T2 Mapping MRI for Prostate Cancer Detection and Classification.

BACKGROUND: Currently, multi-parametric prostate MRI (mpMRI) consists of a qualitative T2 , diffusion weighted, and dynamic contrast enhanced imaging. Quantification of T2 imaging might further standardize PCa detection and support artificial intelligence solutions. PURPOSE: To evaluate the value of T2 mapping to detect prostate cancer (PCa) and to differentiate PCa aggressiveness. STUDY TYPE: Retrospective single center cohort study. POPULATION: Forty-four consecutive patients (mean age 67 years; median PSA 7.9 ng/mL) with mpMRI and verified PCa by subsequent targeted plus systematic MR/ultrasound (US)-fusion biopsy from February 2019 to December 2019. FIELD STRENGTH/SEQUENCE: Standardized mpMRI at 3 T with an additionally acquired T2 mapping sequence. ASSESSMENT: Primary endpoint was the analysis of quantitative T2 values and contrast differences/ratios (CD/CR) between PCa and benign tissue. Secondary objectives were the correlation between T2 values, ISUP grade, apparent diffusion coefficient (ADC) value, and PI-RADS, and the evaluation of thresholds for differentiating PCa and clinically significant PCa (csPCa). STATISTICAL TESTS: Mann-Whitney test, Spearman's rank (rs ) correlation, receiver operating curves, Youden's index (J), and AUC were performed. Statistical significance was defined as P < 0.05. RESULTS: Median quantitative T2 values were significantly lower for PCa in PZ (85 msec) and PCa in TZ (75 msec) compared to benign PZ (141 msec) or TZ (97 msec) (P < 0.001). CD/CR between PCa and benign PZ (51.2/1.77), respectively TZ (19.8/1.29), differed significantly (P < 0.001). The best T2 -mapping threshold for PCa/csPCa detection was for TZ 81/86 msec (J = 0.929/1.0), and for PZ 110 msec (J = 0.834/0.905). Quantitative T2 values of PCa did not correlate significantly with the ISUP grade (rs  = 0.186; P = 0.226), ADC value (rs  = 0.138; P = 0.372), or PI-RADS (rs  = 0.132; P = 0.392). DATA CONCLUSION: Quantitative T2 values could differentiate PCa in TZ and PZ and might support standardization of mpMRI of the prostate. Different thresholds seem to apply for PZ and TZ lesions. However, in the present study quantitative T2 values were not able to indicate PCa aggressiveness. LEVEL OF EVIDENCE: 2 TECHNICAL EFFICACY: Stage 2.

Magnetic resonance cholangiopancreatography with compressed sensing at 1.5 T: clinical application for the evaluation of branch duct IPMN of the pancreas.

OBJECTIVES: To evaluate magnetic resonance cholangiopancreatography (MRCP) with compressed sensing (CS) for the assessment of branch duct intraductal papillary mucinous neoplasm (BD-IPMN) of the pancreas. For this purpose, conventional navigator-triggered (NT) sampling perfection with application-optimized contrast using different flip angle evolutions (SPACE) MRCP was compared with various CS-SPACE-MRCP sequences in a clinical setting. METHODS: A total of 41 patients (14 male, 27 female, mean age 68 years) underwent 1.5-T MRCP for the evaluation of BD-IPMN. The MRCP protocol consisted of the following sequences: conventional NT-SPACE-MRCP, CS-SPACE-MRCP with long (BHL, 17 s) and short single breath-hold (BHS, 8 s), and NT-CS-SPACE-MRCP. Two board-certified radiologists evaluated image quality, duct sharpness, duct visualization, lesion conspicuity, confidence, and communication with the main pancreatic duct in consensus using a 5-point scale (1-5), with higher scores indicating better quality/delineation/confidence. Maximum intensity projection reconstructions and originally acquired data were used for evaluation. Wilcoxon signed-rank test was used to compare the intra-individual difference between sequences. RESULTS: BHS-CS-SPACE-MRCP had the highest scores for image quality (3.85 ± 0.79), duct sharpness (3.81 ± 1.05), and duct visualization (3.81 ± 1.01). There was a significant difference compared with NT-CS-SPACE-MRCP (p < 0.05) but no significant difference to the standard NT-SPACE-MRCP (p > 0.05). Concerning diagnostic quality, BHS-CS-SPACE-MRCP had the highest scores in lesion conspicuity (3.95 ± 0.92), confidence (4.12 ± 1.08), and communication (3.8 ± 1.06), significantly higher compared with NT-SPACE-MRCP, BHL-SPACE-MRCP, and NT-CS-SPACE-MRCP (p = <0.05). CONCLUSIONS: MRCP with CS 3D SPACE for the evaluation of BD-IPMN at 1.5 T provides the best results using a short breath-hold sequence. This approach is feasible and an excellent alternative to standard NT 3D MRCP sequences. KEY POINTS: • 1.5-T MRCP with compressed sensing for the evaluation of branch duct IPMN is a feasible method. • Short breath-hold sequences provide the best results for this purpose.

Combination of an ultrafast TWIST-VIBE Dixon sequence protocol and diffusion-weighted imaging into an accurate easily applicable classification tool for masses in breast MRI.

OBJECTIVES: This study aimed to develop a tool for the classification of masses in breast MRI, based on ultrafast TWIST-VIBE Dixon (TVD) dynamic sequences combined with DWI. TVD sequences allow to abbreviate breast MRI protocols, but provide kinetic information only on the contrast wash-in, and because of the lack of the wash-out kinetics, their diagnostic value might be hampered. A special focus of this study was thus to maintain high diagnostic accuracy in lesion classification. MATERIALS AND METHODS: Sixty-one patients who received breast MRI between 02/2014 and 04/2015 were included, with 83 reported lesions (60 malignant). Our institute's standard breast MRI protocol was complemented by an ultrafast TVD sequence. ADC and peak enhancement of the TVD sequences were integrated into a generalised linear model (GLM) for malignancy prediction. For comparison, a second GLM was calculated using ADC and conventional DCE curve type. The resulting GLMs were evaluated for standard diagnostic parameters. For easy application of the GLMs, nomograms were created. RESULTS: The GLM based on peak enhancement of the TVD and ADC was as equally accurate as the GLM based on conventional DCE and ADC, with no significant differences (sensitivity, 93.3%/93.3%; specificity, 91.3%/87.0%; PPV, 96.6%/94.9%; NPV, 84.0%/83.3%; all, p ≥ 0.315). CONCLUSIONS: This study presents a method to integrate ultrafast TVD sequences into a breast MRI protocol, allowing a reduction of the examination time while maintaining diagnostic accuracy. A GLM based on the combination of TVD-derived peak enhancement and ADC provides high diagnostic accuracy, and can be easily applied using a nomogram. KEY POINTS: • Ultrafast TWIST-VIBE Dixon sequence protocols in combination with diffusion-weighted imaging allow to shorten breast MRI examinations, while diagnostic accuracy is maintained. • Integrating peak enhancement from the TWIST-VIBE Dixon sequence and the apparent diffusion coefficient into a generalised linear model provides a comprehensible image evaluation approach. • This approach is further facilitated by nomograms.

Diagnostic performance of initial enhancement analysis using ultra-fast dynamic contrast-enhanced MRI for breast lesions.

OBJECTIVES: To assess the diagnostic value and contribution to BI-RADS categorisation of initial enhancement on ultra-fast DCE-MRI for differentiating malignant and benign breast lesions. METHODS: The institutional review board approved this study, and written informed consent was obtained from each participant. Both ultra-fast DCE-MRI for initial enhancement analysis and conventional MRI were performed on 200 subjects with a total of 215 lesions (147 malignant and 68 benign). BI-RADS categorisation of enhancing lesions was performed using the conventional MRI. Two initial enhancement measures, time to enhancement (TTE) and maximum slope (MS), were derived from the ultra-fast DCE-MRI. Diagnostic performance and the additional diagnostic value of adding TTE and MS to BI-RADS were evaluated. RESULTS: Both TTE and MS showed significant differences between malignant and benign breast lesions in masses (TTE, p <.001; MS, p = .006) and non-mass enhancement (NME) (TTE, p <.001; MS, p <.001). For masses, the AUC of TTE+MS combined with BI-RADS (0.864) was better than BI-RADS alone (0.823, p = .065). For NME, the AUC of TTE+MS combined with BI-RADS (0.923) was significantly larger than BI-RADS alone (0.865, p = .036), and diagnostic specificity improved by 40.9% (p = .005), without a significant decrease in the sensitivity (p = .083). CONCLUSION: Initial enhancement analysis using ultra-fast DCE-MRI is especially useful for increasing the diagnostic performance of NME in breast MRI. KEY POINTS: • Ultra-fast dynamic MRI effectively differentiates benign from malignant breast lesions. • Ultra-fast dynamic MRI contributes to BI-RADS categorisation in non-mass enhancement. • Management of non-mass breast lesions becomes more appropriate.

Accelerated T2 mapping combining parallel MRI and model-based reconstruction: GRAPPATINI.

BACKGROUND: Quantitative T2 measurements are sensitive to intra- and extracellular water accumulation and myelin loss. Therefore, quantitative T2 promises to be a good biomarker of disease. However, T2 measurements require long acquisition times. PURPOSE: To accelerate T2 quantification and subsequent generation of synthetic T2 -weighted (T2 -w) image contrast for clinical research and routine. To that end, a recently developed model-based approach for rapid T2 and M0 quantification (MARTINI) based on undersampling k-space, was extended by parallel imaging (GRAPPA) to enable high-resolution T2 mapping with access to T2 -w images in less than 2 minutes acquisition time for the entire brain. STUDY TYPE: Prospective cross-sectional study. SUBJECTS/PHANTOM: Fourteen healthy subjects and a multipurpose phantom. FIELD STRENGTH/SEQUENCE: Carr-Purcell-Meiboom-Gill sequence at a 3T scanner. ASSESSMENT: The accuracy and reproducibility of the accelerated T2 quantification was assessed. Validations comprised MRI studies on a phantom as well as the brain, knee, prostate, and liver from healthy volunteers. Synthetic T2 -w images were generated from computed T2 and M0 maps and compared to conventional fast spin-echo (SE) images. STATISTICAL TESTS: Root mean square distance (RMSD) to the reference method and region of interest analysis. RESULTS: The combination of MARTINI and GRAPPA (GRAPPATINI) lead to a 10-fold accelerated T2 mapping protocol with 1:44 minutes acquisition time and full brain coverage. The RMSD of GRAPPATINI increases less (4.3%) than a 10-fold MARTINI reconstruction (37.6%) in comparison to the reference. Reproducibility tests showed low standard deviation (SD) of T2 values in regions of interest between scan and rescan (<0.4 msec) and across subjects (<4 msec). DATA CONCLUSION: GRAPPATINI provides highly reproducible and fast whole-brain T2 maps and arbitrary synthetic T2 -w images in clinically compatible acquisition times of less than 2 minutes. These abilities are expected to support more widespread clinical applications of quantitative T2 mapping. LEVEL OF EVIDENCE: 2 Technical Efficacy: Stage 1 J. MAGN. RESON. IMAGING 2018;48:359-368.

Modified breath-hold compressed-sensing 3D MR cholangiopancreatography with a small field-of-view and high resolution acquisition: Clinical feasibility in biliary and pancreatic disorders.

BACKGROUND: Compressed-sensing (CS) accelerated 3D MR cholangiopancreatography (MRCP) could be acquired in both navigator-triggered (NT) and breath-hold (BH) mode, but the latter has been considered inferior in depicting pancreatic duct and diagnosing pancreatic duct-related diseases. PURPOSE: To prospectively evaluate the clinical feasibility of a modified 3D BH-CS-MRCP prototype protocol with small field-of-view (FOV) and higher spatial resolution, and to compare its performance to the original BH-CS-MRCP and NT-CS-MRCP. STUDY TYPE: Prospective cohort study. POPULATION: Eighty-two patients with suspected pancreaticobiliary diseases (46 male, median age, 55 years, range, 16-79 years), including seven noncooperative patients. FIELD STRENGTH/SEQUENCE: 3T, CS-MRCP. ASSESSMENT: Three protocols were performed in random order in each patient. Acquisition time of each protocol was recorded. Image quality, background suppression, duct visibility, and diagnostic confidence with duct anatomic variations and duct-related pathologies were rated on a 5-point scale by two blinded radiologists independently. STATISTICAL TESTS: The Wilcoxon signed-rank test was used to compare the intraindividual difference. Interobserver agreement was determined using kappa coefficients. The diagnostic performance was calculated using receiver operating characteristic curves. RESULTS: Acquisition time was 17 seconds for both BH-CS-MRCP protocols, and 127.5 ± 36.9 seconds for NT-CS-MRCP. In 75 cooperative patients, the incidence of major artifacts was low for all protocols (5.3-8.0%). Background suppression was similar with the two BH-CS-MRCP protocols (3.67 ± 0.77 for original BH-CS-MRCP and 3.70 ± 0. 57 for modified BH-CS-MRCP, respectively), both inferior to the NT-CS-MRCP protocol (4.41 ± 0.68, P < 0.001 for both). Modified BH-CS-MRCP and NT-CS-MRCP depicted pancreatic duct and second-level branches of biliary duct better than original BH-CS-MRCP (all P < 0.01). The diagnostic performance for detecting bile duct abnormalities was similar for all protocols (P = 0.53-0.87), whereas for detecting pancreatic duct abnormalities, modified BH-CS-MRCP and NT-CS-MRCP had significantly better performance compared to original BH-CS-MRCP (both P < 0.01). In seven noncooperative patients, NT-CS-MRCP had superior image quality than both BH protocols (both P < 0.01). DATA CONCLUSION: Modified BH-CS-MRCP is feasible for pancreatic and biliary disorders. NT-CS-MRCP might be more useful in noncooperative patients. LEVEL OF EVIDENCE: 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;47:1389-1399.

Flexible proton 3D MR spectroscopic imaging of the prostate with low-power adiabatic pulses for volume selection and spiral readout.

PURPOSE: Cartesian k-space sampling in three-dimensional magnetic resonance spectroscopic imaging (MRSI) of the prostate limits the selection of voxel size and acquisition time. Therefore, large prostates are often scanned at reduced spatial resolutions to stay within clinically acceptable measurement times. Here we present a semilocalized adiabatic selective refocusing (sLASER) sequence with gradient-modulated offset-independent adiabatic (GOIA) refocusing pulses and spiral k-space acquisition (GOIA-sLASER-Spiral) for fast prostate MRSI with enhanced resolution and extended matrix sizes. METHODS: MR was performed at 3 tesla with an endorectal receive coil. GOIA-sLASER-Spiral at an echo time (TE) of 90 ms was compared to a point-resolved spectroscopy sequence (PRESS) with weighted, elliptical phase encoding at an TE of 145 ms using simulations and measurements of phantoms and patients (n = 9). RESULTS: GOIA-sLASER-Spiral acquisition allows prostate MR spectra to be obtained in ∼5 min with a quality comparable to those acquired with a common Cartesian PRESS protocol in ∼9 min, or at an enhanced spatial resolution showing more precise tissue allocation of metabolites. Extended field of views (FOVs) and matrix sizes for large prostates are possible without compromising spatial resolution or measurement time. CONCLUSION: The flexibility of spiral sampling enables prostate MRSI with a wide range of resolutions and FOVs without undesirable increases in acquisition times, as in Cartesian encoding. This approach is suitable for routine clinical exams of prostate metabolites. Magn Reson Med 77:928-935, 2017. © 2016 International Society for Magnetic Resonance in Medicine.

Metabolite ratios in 1H MR spectroscopic imaging of the prostate.

In (1)H MR spectroscopic imaging ((1)H-MRSI) of the prostate the spatial distribution of the signal levels of the metabolites choline, creatine, polyamines, and citrate are assessed. The ratio of choline (plus spermine as the main polyamine) plus creatine over citrate [(Cho+(Spm+)Cr)/Cit] is derived from these metabolites and is used as a marker for the presence of prostate cancer. In this review, the factors that are of importance for the metabolite ratio are discussed. This is relevant, because the appearance of the metabolites in the spectrum depends not only on the underlying anatomy, metabolism, and physiology of the tissue, but also on acquisition parameters. These parameters influence especially the spectral shapes of citrate and spermine resonances, and consequently, the (Cho+(Spm+)Cr)/Cit ratio. Both qualitative and quantitative approaches can be used for the evaluation of (1)H-MRSI spectra of the prostate. For the quantitative approach, the (Cho+(Spm+)Cr)/Cit ratio can be determined by integration or by a fit based on model signals. Using the latter, the influence of the acquisition parameters on citrate can be taken into account. The strong overlap between the choline, creatine, and spermine resonances complicates fitting of the individual metabolites. This overlap and (unknown, possibly tissue-related) variations in T1, T2, and J-modulation hamper the application of corrections needed for a "normalized" (Cho+(Spm+)Cr)/Cit ratio that would enable comparison of spectra measured with different prostate MR spectroscopy protocols. Quantitative (Cho+(Spm+)Cr)/Cit thresholds for the evaluation of prostate cancer are therefore commonly established per institution or per protocol. However, if the same acquisition and postprocessing protocol were used, the ratio and the thresholds would be institution-independent, promoting the clinical usability of prostate (1)H-MRSI.

Improved volume selective (1) H MR spectroscopic imaging of the prostate with gradient offset independent adiabaticity pulses at 3 tesla.

PURPOSE: Volume selection in (1) H MR spectroscopic imaging (MRSI) of the prostate is commonly performed with low-bandwidth refocusing pulses. However, their large chemical shift displacement error (CSDE) causes lipid signal contamination in the spectral range of interest. Application of high-bandwidth adiabatic pulses is limited by radiofrequency (RF) power deposition. In this study, we aimed to provide an MRSI sequence that overcomes these limitations. METHODS: Measurements were performed at 3 T with an endorectal receive coil. A semi-LASER sequence was equipped with low RF power demanding gradient-modulated offset-independent adiabaticity (GOIA) refocusing pulses with WURST(16,4) modulation, with a 10 kHz bandwidth. RESULTS: Simulations and phantom studies verified that the GOIA pulses select slices with a flat top and sharp edges and minimal CSDE. The sequence timing was tuned to an optimal citrate signal shape at an echo time of 88 ms. Patient studies (n = 10) demonstrated that high quality spectra with reduced lipid artifacts can be obtained from the whole prostate. Compared with PRESS acquisition at 145 ms the signal-to-noise ratio (SNR) of citrate is increased up to 2.6 and choline up to 1.3. CONCLUSION: An MRSI sequence of the prostate is presented with minimized spectral lipid contamination and improved SNR, to facilitate routine clinical acquisition of metabolic data.

(31) P MR spectroscopic imaging of the human prostate at 7 T: T1 relaxation times, Nuclear Overhauser Effect, and spectral characterization.

PURPOSE: Optimization of phosphorus ((31) P) MR spectroscopic imaging (MRSI) of the human prostate at 7 T by the evaluation of T1 relaxation times and the Nuclear Overhauser Effect (NOE) of phosphorus-containing metabolites. METHODS: Twelve patients with prostate cancer and one healthy volunteer were scanned on a 7 T whole-body system using a (31) P endorectal coil combined with an eight-channel (1) H body array coil. T1 relaxation times were measured using progressive saturation in a two-dimensional localization sequence. (31) P MRSI was performed twice: once without NOE and once with NOE using low-power continuous wave (1) H irradiation to determine NOE enhancements. RESULTS: T1 relaxation times of (31) P metabolites in the human prostate at 7 T varied between 3.0 and 8.3 s. Positive but variable NOE enhancements were measured for most metabolites. Remarkably, the (31) P MR spectra showed two peaks in chemical shift range of inorganic phosphate. CONCLUSION: Knowledge of T1 relaxation times and NOE enhancements enables protocol optimization for (31) P MRSI of the prostate at 7 T. With a strongly reduced (31) P flip angle (≤ 45°), a (31) P MRSI dataset with optimal signal-to-noise ratio per unit time can be obtained within 15 minutes. The NOE enhancement can improve fitting accuracy, but its variability requires further investigation.

Optimizing Image Quality with High-Resolution, Deep-Learning-Based Diffusion-Weighted Imaging in Breast Cancer Patients at 1.5 T.

The objective of this study was to evaluate a high-resolution deep-learning (DL)-based diffusion-weighted imaging (DWI) sequence for breast magnetic resonance imaging (MRI) in comparison to a standard DWI sequence (DWIStd) at 1.5 T. It is a prospective study of 38 breast cancer patients, who were scanned with DWIStd and DWIDL. Both DWI sequences were scored for image quality, sharpness, artifacts, contrast, noise, and diagnostic confidence with a Likert-scale from 1 (non-diagnostic) to 5 (excellent). The lesion diameter was evaluated on b 800 DWI, apparent diffusion coefficient (ADC), and the second subtraction (SUB) of the contrast-enhanced T1 VIBE. SNR was also calculated. Statistics included correlation analyses and paired t-tests. High-resolution DWIDL offered significantly superior image quality, sharpness, noise, contrast, and diagnostic confidence (each p < 0.02)). Artifacts were significantly higher in DWIDL by one reader (M = 4.62 vs. 4.36 Likert scale, p < 0.01) without affecting the diagnostic confidence. SNR was higher in DWIDL for b 50 and ADC maps (each p = 0.07). Acquisition time was reduced by 22% in DWIDL. The lesion diameters in DWI b 800DL and Std and ADCDL and Std were respectively 6% lower compared to the 2nd SUB. A DL-based diffusion sequence at 1.5 T in breast MRI offers a higher resolution and a faster acquisition, including only minimally more artefacts without affecting the diagnostic confidence.

Shortening Acquisition Time and Improving Image Quality for Pelvic MRI Using Deep Learning Reconstruction for Diffusion-Weighted Imaging at 1.5 T.

RATIONALE AND OBJECTIVES: To determine the impact on acquisition time reduction and image quality of a deep learning (DL) reconstruction for accelerated diffusion-weighted imaging (DWI) of the pelvis at 1.5 T compared to standard DWI. MATERIALS AND METHODS: A total of 55 patients (mean age, 61 ± 13 years; range, 27-89; 20 men, 35 women) were consecutively included in this retrospective, monocentric study between February and November 2022. Inclusion criteria were (1) standard DWI (DWIS) in clinically indicated magnetic resonance imaging (MRI) at 1.5 T and (2) DL-reconstructed DWI (DWIDL). All patients were examined using the institution's standard MRI protocol according to their diagnosis including DWI with two different b-values (0 and 800 s/mm2) and calculation of apparent diffusion coefficient (ADC) maps. Image quality was qualitatively assessed by four radiologists using a visual 5-point Likert scale (5 = best) for the following criteria: overall image quality, noise level, extent of artifacts, sharpness, and diagnostic confidence. The qualitative scores for DWIS and DWIDL were compared with the Wilcoxon signed-rank test. RESULTS: The overall image quality was evaluated to be significantly superior in DWIDL compared to DWIS for b = 0 s/mm2, b = 800 s/mm2, and ADC maps by all readers (P < .05). The extent of noise was evaluated to be significantly less in DWIDL compared to DWIS for b = 0 s/mm2, b = 800 s/mm2, and ADC maps by all readers (P < .001). No significant differences were found regarding artifacts, lesion detectability, sharpness of organs, and diagnostic confidence (P > .05). Acquisition time for DWIS was 2:06 minutes, and simulated acquisition time for DWIDL was 1:12 minutes. CONCLUSION: DL image reconstruction improves image quality, and simulation results suggest that a reduction in acquisition time for diffusion-weighted MRI of the pelvis at 1.5 T is possible.

AI-Supported Autonomous Uterus Reconstructions: First Application in MRI Using 3D SPACE with Iterative Denoising.

RATIONALE AND OBJECTIVES: T2-weighted imaging in at least two orthogonal planes is recommended for assessment of the uterus. To determine whether a convolutional neural network-based algorithm could be used for the re-constructions of uterus axes derived from a 3D SPACE with iterative denoising. MATERIALS AND METHODS: 50 patients aged 18-81 (mean: 42) years who underwent an MRI examination of the uterus participated voluntarily in this prospective study after informed consent. In addition to a standard MRI pelvis protocol, a 3D SPACE research application sequence was acquired in sagittal orientation. Reconstructions for both the cervix and the cavum in the short and long axes were performed by a research trainee (T), an experienced radiologist (E), and the prototype software (P). In the next step, the reconstructions were evaluated anonymously by two experienced readers according to 5-point-Likert-Scales. In addition, the length of the cervical canal, the length of the cavum and the distance between the tube angles were measured on all reconstructions. Interobserver agreement was assessed for all ratings. RESULTS: For all axes, significant differences were found between the scores of the reconstructions by research T, E and P. P received higher scores and was preferred significantly more often with the exception of the comparison of the reconstruction Cervix short of E (Cervix short: P vs. T: p = 0.02; P vs. E: p = 0.26; Cervix long: P vs. T: p = 0.01; P vs. E: p < 0.01; Cavum short: P vs. T: p = 0.01; P vs. E: p = 0.02; Cavum long: P vs. T: p < 0.01; P vs. E: p < 0.01). Regarding the measured diameters, (length of cervical canal/cavum/distance between tube angles) significantly larger diameters were recorded for P compared to E and T (Cervix long (mm): T: 25.43; E: 25.65; P: 26.65; Cavum short (mm): T: 26.24; E: 25.04; P: 27.33; Cavum long (mm): T: 31.98; E: 32.91; P: 34.41; P vs. T: p < 0.01); P vs. E: p = 0.04). Moderate to substantial agreement was found between Reader 1 and Reader 2 (range: 0.39-0.67). CONCLUSION: P was able to reconstruct the axes at least as well as or better than E and T. P could thereby lead to workflow facilitation and enable more efficient reporting of uterine MRI.

Stability of Radiomic Features against Variations in Lesion Segmentations Computed on Apparent Diffusion Coefficient Maps of Breast Lesions.

Diffusion-weighted imaging (DWI) combined with radiomics can aid in the differentiation of breast lesions. Segmentation characteristics, however, might influence radiomic features. To evaluate feature stability, we implemented a standardized pipeline featuring shifts and shape variations of the underlying segmentations. A total of 103 patients were retrospectively included in this IRB-approved study after multiparametric diagnostic breast 3T MRI with a spin-echo diffusion-weighted sequence with echoplanar readout (b-values: 50, 750 and 1500 s/mm2). Lesion segmentations underwent shifts and shape variations, with >100 radiomic features extracted from apparent diffusion coefficient (ADC) maps for each variation. These features were then compared and ranked based on their stability, measured by the Overall Concordance Correlation Coefficient (OCCC) and Dynamic Range (DR). Results showed variation in feature robustness to segmentation changes. The most stable features, excluding shape-related features, were FO (Mean, Median, RootMeanSquared), GLDM (DependenceNonUniformity), GLRLM (RunLengthNonUniformity), and GLSZM (SizeZoneNonUniformity), which all had OCCC and DR > 0.95 for both shifting and resizing the segmentation. Perimeter, MajorAxisLength, MaximumDiameter, PixelSurface, MeshSurface, and MinorAxisLength were the most stable features in the Shape category with OCCC and DR > 0.95 for resizing. Considering the variability in radiomic feature stability against segmentation variations is relevant when interpreting radiomic analysis of breast DWI data.

Enhancing diffusion-weighted prostate MRI through self-supervised denoising and evaluation.

Diffusion-weighted imaging (DWI) is a magnetic resonance imaging (MRI) technique that provides information about the Brownian motion of water molecules within biological tissues. DWI plays a crucial role in stroke imaging and oncology, but its diagnostic value can be compromised by the inherently low signal-to-noise ratio (SNR). Conventional supervised deep learning-based denoising techniques encounter challenges in this domain as they necessitate noise-free target images for training. This work presents a novel approach for denoising and evaluating DWI scans in a self-supervised manner, eliminating the need for ground-truth data. By leveraging an adapted version of Stein's unbiased risk estimator (SURE) and exploiting a phase-corrected combination of repeated acquisitions, we outperform both state-of-the-art self-supervised denoising methods and conventional non-learning-based approaches. Additionally, we demonstrate the applicability of our proposed approach in accelerating DWI scans by acquiring fewer image repetitions. To evaluate denoising performance, we introduce a self-supervised methodology that relies on analyzing the characteristics of the residual signal removed by the denoising approaches.

Deep learning image reconstruction of diffusion-weighted imaging in evaluation of prostate cancer focusing on its clinical implications.

Background: Image-based assessment of prostate cancer (PCa) is increasingly emphasized in the diagnostic workflow for selecting biopsy targets and possibly predicting clinically significant prostate cancer (csPCa). Assessment is based on Prostate Imaging-Reporting and Data System (PI-RADS) which is largely dependent on T2-weighted image (T2WI) and diffusion weighted image (DWI). This study aims to determine whether deep learning reconstruction (DLR) can improve the image quality of DWI and affect the assessment of PI-RADS ≥4 in patients with PCa. Methods: In this retrospective study, 3.0T post-biopsy prostate magnetic resonance imaging (MRI) of 70 patients with PCa in Korea University Ansan Hospital from November 2021 to July 2022 was reconstructed with and without using DLR. Four DWI image sets were made: (I) conventional DWI (CDWI): DWI with acceleration factor 2 and conventional parallel imaging reconstruction, (II) DL1: DWI with acceleration factor 2 using DLR, (III) DL2: DWI with acceleration factor 3 using DLR, and (IV) DL3: DWI with acceleration factor 3 and half average b-value using DLR. Apparent diffusion coefficient (ADC) value, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were measured by one reviewer, while two reviewers independently assessed overall image quality, noise, and lesion conspicuity using a four-point visual scoring system from each DWI image set. Two reviewers also performed PI-RADSv2.1 scoring on lesions suspected of malignancy. Results: A total of 70 patients (mean age, 70.8±9.7 years) were analyzed. The image acquisition time was 4:46 min for CDWI and DL1, 3:40 min for DL2, and 2:00 min for DL3. DL1 and DL2 images resulted in better lesion conspicuity compared to CDWI images assessed by both readers (P<0.05). DLR resulted in a significant increase in SNR, from 38.4±14.7 in CDWI to 56.9±21.0 in DL1. CNR increased from 25.1±11.5 in CDWI to 43.1±17.8 in DL1 (P<0.001). PI-RADS v2.1 scoring for PCa lesions was more agreeable with the DL1 reconstruction method than with CDWI (κ value CDWI, DL1; 0.40, 0.61, respectively). A statistically significant number of lesions were upgraded from PI-RADS <4 in CDWI image to PI-RADS ≥4 in DL1 images for both readers (P<0.05). Most of the PI-RADS upgraded lesions were from higher than unfavorable intermediate-risk groups according to the 2023 National Comprehensive Cancer Network guidelines with statistically significant difference of marginal probability in DL1 and DL2 for both readers (P<0.05). Conclusions: DLR in DWI for PCa can provide options for improving image quality with a significant impact on PI-RADS evaluation or about a 23% reduction in acquisition time without compromising image quality.

Thin-Slice Prostate MRI Enabled by Deep Learning Image Reconstruction.

OBJECTIVES: Thin-slice prostate MRI might be beneficial for prostate cancer diagnostics. However, prolongation of acquisition time is a major drawback of thin-slice imaging. Therefore, the purpose of this study was to investigate the impact of a thin-slice deep learning accelerated T2-weighted (w) TSE imaging sequence (T2DLR) of the prostate as compared to conventional T2w TSE imaging (T2S). MATERIALS AND METHODS: Thirty patients were included in this prospective study at one university center after obtaining written informed consent. T2S (3 mm slice thickness) was acquired first in three orthogonal planes followed by thin-slice T2DLR (2 mm slice thickness) in axial plane. Acquisition time of axial conventional T2S was 4:12 min compared to 4:37 min for T2DLR. Imaging datasets were evaluated by two radiologists using a Likert-scale ranging from 1-4, with 4 being the best regarding the following parameters: sharpness, lesion detectability, artifacts, overall image quality, and diagnostic confidence. Furthermore, preference of T2S versus T2DLR was evaluated. RESULTS: The mean patient age was 68 ± 8 years. Sharpness of images and lesion detectability were rated better in T2DLR with a median of 4 versus a median of 3 in T2S (p < 0.001 for both readers). Image noise was evaluated to be significantly worse in T2DLR as compared to T2S (p < 0.001 and p = 0.021, respectively). Overall image quality was also evaluated to be superior in T2DLR versus T2S with a median of 4 versus 3 (p < 0.001 for both readers). Both readers chose T2DLR in 29 cases as their preference. CONCLUSIONS: Thin-slice T2DLR of the prostate provides a significant improvement of image quality without significant prolongation of acquisition time.

Acquisition time reduction of diffusion-weighted liver imaging using deep learning image reconstruction.

PURPOSE: The purpose of this study was to investigate the impact of deep learning accelerated diffusion-weighted imaging (DWIDL) in 1.5-T liver MRI on image quality, sharpness, and diagnostic confidence. MATERIALS AND METHODS: One-hundred patients who underwent liver MRI at 1.5-T including DWI with two different b-values (50 and 800 s/mm²) between February and April 2022 were retrospectively included. There were 54 men and 46 women, with a mean age of 59 ± 14 (SD) years (range: 21-88 years). The single average raw data were retrospectively processed using a deep learning (DL) image reconstruction algorithm leading to a simulated acquisition time of 1 min 28 s for DWIDL as compared to 2 min 31 s for standard DWI (DWIStd) via reduction of signal averages. All DWI datasets were reviewed by four radiologists using a Likert scale ranging from 1-4 using the following criteria: noise level, extent of artifacts, sharpness, overall image quality, and diagnostic confidence. Furthermore, quantitative assessment of noise and signal-to-noise ratio (SNR) was performed via regions of interest. RESULTS: No significant differences were found regarding artifacts and overall image quality (P > 0.05). Noise measurements for the spleen, liver, and erector spinae muscles revealed significantly lower noise for DWIDL versus DWIStd (P < 0.001). SNR measurements in the above-mentioned tissues also showed significantly superior results for DWIDL versus DWIStd for b = 50 s/mm² and ADC maps (all P < 0.001). For b = 800 s/mm², significantly superior results were found for the spleen, right hemiliver, and erector spinae muscles. CONCLUSIONS: DL image reconstruction of liver DWI at 1.5-T is feasible including significant reduction of acquisition time without compromised image quality.