Repeatability of diffusion-weighted MRI of the prostate using whole lesion ADC values, skew and histogram analysis

Feasibility of ADC histogram analysis for predicting of postoperative recurrence in aggressive spinal tumors

Background

Risk stratification of spinal tumors is a major unmet clinical need for personalized therapy.

Purpose

To explore the feasibility of pretreatment whole-lesion apparent diffusion coefficient (ADC) histogram in predicting local recurrence of aggressive spinal tumors.

Methods

119 aggressive spinal tumor patients (median age, 40; range, 13-74  years) confirmed by pathological findings with a mean follow-up of 36 months were enrolled and divided into the recurrence and non-recurrence group. The histogram metrics of whole-lesion, including the maximum, mean, kurtosis, skewness, entropy, and percentiles (10th, 25th, 50th, 75th, 95th) ADC values, were evaluated and take the average. Fractal dimension (FD) was assessed in the three orthogonal directions and take maximum. Clinical and general imaging features were used to construct an alternative prognostic model for comparison. Variables with statistical differences would be included in stepwise logistic regression analysis.

Results

As for the clinical model, Enneking staging (odds ratio [OR]: 3.572; P = 0.04) and vertebral compression (OR: 4.302; P = 0.002) were independent predictors of recurrence. There was no statistical difference in FD between the two groups (P = 0.623). Among the ADC histogram parameters compared, skewness, maximum, and mean ADC values were independent risk factors and constructed ADC histogram prediction models. The ADC histogram model (AUC = 0.871) and the combined model (AUC = 0.884) performed better than the clinical prediction model (AUC = 0.704) with P-values of 0.004 and 0.001, respectively.

Conclusion

Prediction models based on the ADC histogram analysis might represent serviceable instruments for the aggressive spinal tumors.

Anatomical zone and tissue type impacts the repeatability of quantitative MRI parameters and radiomic features for longitudinal monitoring of treatment response in the prostate

OBJECTIVE

To (1) establish the repeatability coefficient (%RC) of region of interest (ROI) and voxel-wise measurements of a comprehensive range of quantitative MRI (qMRI) parameters and radiomic features in the prostate, and (2) assess the impact of different tissue types (benign vs tumor) and anatomical zones (peripheral, PZ, and non-peripheral, nPZ) on the %RCs.

METHODS

Test-retest qMRI was acquired in ten prostate cancer patients and six healthy volunteers. Parametric maps of apparent diffusion coefficient (ADC), diffusion coefficient (D), perfusion fraction (f), hypoxia score (HS), longitudinal relaxation time (T1), and observed transverse relaxation rate (R2*) were calculated. Fifty-nine radiomic feature maps were calculated from each of the parametric maps and T2-weighted images. The %RCs between tissue type and anatomical zones were compared using the Student's t test at 95% significance level.

RESULTS

The %RC of ADC, D and HS, and up to 118 (out of all 413) radiomic features was significantly different between either anatomical zones, or between tumor and benign tissue, or both.

CONCLUSIONS

DWI-derived parameters and a portion of their radiomic features require %RCs to be established specifically for anatomical zones, tumor and benign tissues. The remaining qMRI parameters and features can have a single threshold for the whole prostate.

The QIBA Profile for Diffusion-Weighted MRI: Apparent Diffusion Coefficient as a Quantitative Imaging Biomarker

The apparent diffusion coefficient (ADC) provides a quantitative measure of water mobility that can be used to probe alterations in tissue microstructure due to disease or treatment. Establishment of the accepted level of variance in ADC measurements for each clinical application is critical for its successful implementation. The Diffusion-Weighted Imaging Biomarker Committee of the Quantitative Imaging Biomarkers Alliance (QIBA) has recently advanced the ADC Profile from the consensus to clinically feasible stage for the brain, liver, prostate, and breast. This profile distills multiple studies on ADC repeatability and describes detailed procedures to achieve stated performance claims on an observed ADC change within acceptable confidence limits. In addition to reviewing the current ADC Profile claims, this report has used recent literature to develop proposed updates for establishing metrology benchmarks for mean lesion ADC change that account for measurement variance. Specifically, changes in mean ADC exceeding 8% for brain lesions, 27% for liver lesions, 27% for prostate lesions, and 15% for breast lesions are claimed to represent true changes with 95% confidence. This report also discusses the development of the ADC Profile, highlighting its various stages, and describes the workflow essential to achieving a standardized implementation of advanced quantitative diffusion-weighted MRI in the clinic. The presented QIBA ADC Profile guidelines should enable successful clinical application of ADC as a quantitative imaging biomarker and ensure reproducible ADC measurements that can be used to confidently evaluate longitudinal changes and treatment response for individual patients.

Deep Learning Prostate MRI Segmentation Accuracy and Robustness: A Systematic Review

Purpose To investigate the accuracy and robustness of prostate segmentation using deep learning across various training data sizes, MRI vendors, prostate zones, and testing methods relative to fellowship-trained diagnostic radiologists. Materials and Methods In this systematic review, Embase, PubMed, Scopus, and Web of Science databases were queried for English-language articles using keywords and related terms for prostate MRI segmentation and deep learning algorithms dated to July 31, 2022. A total of 691 articles from the search query were collected and subsequently filtered to 48 on the basis of predefined inclusion and exclusion criteria. Multiple characteristics were extracted from selected studies, such as deep learning algorithm performance, MRI vendor, and training dataset features. The primary outcome was comparison of mean Dice similarity coefficient (DSC) for prostate segmentation for deep learning algorithms versus diagnostic radiologists. Results Forty-eight studies were included. Most published deep learning algorithms for whole prostate gland segmentation (39 of 42 [93%]) had a DSC at or above expert level (DSC ≥ 0.86). The mean DSC was 0.79 ± 0.06 (SD) for peripheral zone, 0.87 ± 0.05 for transition zone, and 0.90 ± 0.04 for whole prostate gland segmentation. For selected studies that used one major MRI vendor, the mean DSCs of each were as follows: General Electric (three of 48 studies), 0.92 ± 0.03; Philips (four of 48 studies), 0.92 ± 0.02; and Siemens (six of 48 studies), 0.91 ± 0.03. Conclusion Deep learning algorithms for prostate MRI segmentation demonstrated accuracy similar to that of expert radiologists despite varying parameters; therefore, future research should shift toward evaluating segmentation robustness and patient outcomes across diverse clinical settings. Keywords: MRI, Genital/Reproductive, Prostate Segmentation, Deep Learning Systematic review registration link: osf.io/nxaev © RSNA, 2024.

Fat Fraction Extracted from Whole-Body Magnetic Resonance (WB-MR) in Bone Metastatic Prostate Cancer: Intra- and Inter-Reader Agreement of Single-Slice and Volumetric Measurements

BACKGROUND

This study evaluates the repeatability and reproducibility of fat-fraction percentage (FF%) in whole-body magnetic resonance imaging (WB-MRI) of prostate cancer patients with bone metastatic hormone naive disease.

METHODS

Patients were selected from the database of a prospective phase-II trial. The treatment response was assessed using the METastasis Reporting and Data System for Prostate (MET-RADS-P). Two operators identified a Small Active Lesion (SAL, <10 mm) and a Large Active Lesion (LAL, ≥10 mm) per patient, performing manual segmentation of lesion volume and the largest cross-sectional area. Measurements were repeated by one operator after two weeks. Intra- and inter-reader agreements were assessed via Interclass Correlation Coefficient (ICC) on first-order radiomics features.

RESULTS

Intra-reader ICC showed high repeatability for both SAL and LAL in a single slice (SS) and volumetric (VS) measurements with values ranging from 0.897 to 0.971. Inter-reader ICC ranged from 0.641 to 0.883, indicating moderate to good reproducibility. Spearman's rho analysis confirmed a strong correlation between SS and VS measurements for SAL (0.817) and a moderate correlation for LAL (0.649). Both intra- and inter-rater agreement exceeded 0.75 for multiple first-order features across lesion sizes.

CONCLUSION

This study suggests that FF% measurements are reproducible, particularly for larger lesions in both SS and VS assessments.

A statistical framework for planning and analysing test-retest studies of repeatability

There is an increasing number of potential quantitative biomarkers that could allow for early assessment of treatment response or disease progression. However, measurements of such biomarkers are subject to random variability. Hence, differences of a biomarker in longitudinal measurements do not necessarily represent real change but might be caused by this random measurement variability. Before utilizing a quantitative biomarker in longitudinal studies, it is therefore essential to assess the measurement repeatability. Measurement repeatability obtained from test-retest studies can be quantified by the repeatability coefficient, which is then used in the subsequent longitudinal study to determine if a measured difference represents real change or is within the range of expected random measurement variability. The quality of the point estimate of the repeatability coefficient, therefore, directly governs the assessment quality of the longitudinal study. Repeatability coefficient estimation accuracy depends on the case number in the test-retest study, but despite its pivotal role, no comprehensive framework for sample size calculation of test-retest studies exists. To address this issue, we have established such a framework, which allows for flexible sample size calculation of test-retest studies, based upon newly introduced criteria concerning assessment quality in the longitudinal study. This also permits retrospective assessment of prior test-retest studies.

Efficiency and Accuracy Evaluation of Multiple Diffusion-Weighted MRI Techniques Across Different Scanners

BACKGROUND

The choice between different diffusion-weighted imaging (DWI) techniques is difficult as each comes with tradeoffs for efficient clinical routine imaging and apparent diffusion coefficient (ADC) accuracy.

PURPOSE

To quantify signal-to-noise-ratio (SNR) efficiency, ADC accuracy, artifacts, and distortions for different DWI acquisition techniques, coils, and scanners.

STUDY TYPE

Phantom, in vivo intraindividual biomarker accuracy between DWI techniques and independent ratings.

POPULATION/PHANTOMS

NIST diffusion phantom. 51 Patients: 40 with prostate cancer and 11 with head-and-neck cancer at 1.5 T FIELD STRENGTH/SEQUENCE: Echo planar imaging (EPI): 1.5 T and 3 T Siemens; 3 T Philips. Distortion-reducing: RESOLVE (1.5 and 3 T Siemens); Turbo Spin Echo (TSE)-SPLICE (3 T Philips). Small field-of-view (FOV): ZoomitPro (1.5 T Siemens); IRIS (3 T Philips). Head-and-neck and flexible coils.

ASSESSMENT

SNR Efficiency, geometrical distortions, and susceptibility artifacts were quantified for different b-values in a phantom. ADC accuracy/agreement was quantified in phantom and for 51 patients. In vivo image quality was independently rated by four experts.

STATISTICAL TESTS

QIBA methodology for accuracy: trueness, repeatability, reproducibility, Bland-Altman 95% Limits-of-Agreement (LOA) for ADC. Wilcoxon Signed-Rank and student tests on P < 0.05 level.

RESULTS

The ZoomitPro small FOV sequence improved b-image efficiency by 8%-14%, reduced artifacts and observer scoring for most raters at the cost of smaller FOV compared to EPI. The TSE-SPLICE technique reduced artifacts almost completely at a 24% efficiency cost compared to EPI for b-values ≤500 sec/mm2 . Phantom ADC 95% LOA trueness were within ±0.03 × 10-3  mm2 /sec except for small FOV IRIS. The in vivo ADC agreement between techniques, however, resulted in 95% LOAs in the order of ±0.3 × 10-3  mm2 /sec with up to 0.2 × 10-3  mm2 /sec of bias.

DATA CONCLUSION

ZoomitPro for Siemens and TSE SPLICE for Philips resulted in a trade-off between efficiency and artifacts. Phantom ADC quality control largely underestimated in vivo accuracy: significant ADC bias and variability was found between techniques in vivo.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY STAGE: 2.

The PRECISE Recommendations for Prostate MRI in Patients on Active Surveillance for Prostate Cancer: A Critical Review

The Prostate Cancer Radiological Estimation of Change in Sequential Evaluation (PRECISE) recommendations were published in 2016 to standardize the reporting of MRI examinations performed to assess for disease progression in patients on active surveillance for prostate cancer. Although a limited number of studies have reported outcomes from use of PRECISE in clinical practice, the available studies have demonstrated PRECISE to have high pooled NPV but low pooled PPV for predicting progression. Our experience in using PRECISE in clinical practice at two teaching hospitals has highlighted issues with its application and areas requiring clarification. This Clinical Perspective critically appraises PRECISE on the basis of this experience, focusing on the system's key advantages and disadvantages and exploring potential changes to improve the system's utility. These changes include consideration of image quality when applying PRECISE scoring, incorporation of quantitative thresholds for disease progression, adoption of a PRECISE 3F sub-category for progression not qualifying as substantial, and comparisons with both the baseline and most recent prior examinations. Items requiring clarification include derivation of a patient-level score in patients with multiple lesions, intended application of PRECISE score 5 (i.e., if requiring development of disease that is no longer organ-confined), and categorization of new lesions in patients with prior MRI-invisible disease.

Same-day repeatability and Between-Sequence reproducibility of Mean ADC in PI-RADS lesions

PURPOSE

This study aimed to assess repeatability after repositioning (inter-scan), intra-rater, inter-rater and inter-sequence variability of mean apparent diffusion coefficient (ADC) measurements in MRI-detected prostate lesions.

METHOD

Forty-three patients with suspicion for prostate cancer were included and received a clinical prostate bi-/multiparametric MRI examination with repeat scans of the T2-weighted and two DWI-weighted sequences (ssEPI and rsEPI). Two raters (R1 and R2) performed single-slice, 2D regions of interest (2D-ROIs) and 3D-segmentation-ROIs (3D-ROIs). Mean bias, corresponding limits of agreement (LoA), mean absolute difference, within-subject coefficient of variation (CoV) and repeatability/reproducibility coefficient (RC/RDC) were calculated. Bradley & Blackwood test was used for variance comparison. Linear mixed models (LMM) were used to account for multiple lesions per patient.

RESULTS

Inter-scan repeatability, intra-rater and inter-sequence reproducibility analysis of ADC showed no significant bias. 3D-ROIs demonstrated significantly less variability than 2D-ROIs (p < 0.01). Inter-rater comparison demonstrated small significant systematic bias of 57 × 10-6 mm2/s for 3D-ROIs (p < 0.001). Intra-rater RC, with the lowest variation, was 145 and 189 × 10-6 mm2/s for 3D- and 2D-ROIs, respectively. For 3D-ROIs of ssEPI, RCs and RDCs were 190-198 × 10-6 mm2/s for inter-scan, inter-rater and inter-sequence variation. No significant differences were found for inter-scan, inter-rater and inter-sequence variability.

CONCLUSIONS

In a single-scanner setting, single-slice ADC measurements showed considerable variation, which may be lowered using 3D-ROIs. For 3D-ROIs, we propose a cut-off of ∼ 200 × 10-6 mm2/s for differences introduced by repositioning, rater or sequence effects. The results suggest that follow-up measurements should be possible by different raters or sequences.

Machine learning-based radiomics model to predict benign and malignant PI-RADS v2.1 category 3 lesions: a retrospective multi-center study

Purpose

To develop machine learning-based radiomics models derive from different MRI sequences for distinction between benign and malignant PI-RADS 3 lesions before intervention, and to cross-institution validate the generalization ability of the models.

Methods

The pre-biopsy MRI datas of 463 patients classified as PI-RADS 3 lesions were collected from 4 medical institutions retrospectively. 2347 radiomics features were extracted from the VOI of T2WI, DWI and ADC images. The ANOVA feature ranking method and support vector machine classifier were used to construct 3 single-sequence models and 1 integrated model combined with the features of three sequences. All the models were established in the training set and independently verified in the internal test and external validation set. The AUC was used to compared the predictive performance of PSAD with each model. Hosmer–lemeshow test was used to evaluate the degree of fitting between prediction probability and pathological results. Non-inferiority test was used to check generalization performance of the integrated model.

Results

The difference of PSAD between PCa and benign lesions was statistically significant (P = 0.006), with the mean AUC of 0.701 for predicting clinically significant prostate cancer (internal test AUC = 0.709 vs. external validation AUC = 0.692, P = 0.013) and 0.630 for predicting all cancer (internal test AUC = 0.637 vs. external validation AUC = 0.623, P = 0.036). T2WI-model with the mean AUC of 0.717 for predicting csPCa (internal test AUC = 0.738 vs. external validation AUC = 0.695, P = 0.264) and 0.634 for predicting all cancer (internal test AUC = 0.678 vs. external validation AUC = 0.589, P = 0.547). DWI-model with the mean AUC of 0.658 for predicting csPCa (internal test AUC = 0.635 vs. external validation AUC = 0.681, P = 0.086) and 0.655 for predicting all cancer (internal test AUC = 0.712 vs. external validation AUC = 0.598, P = 0.437). ADC-model with the mean AUC of 0.746 for predicting csPCa (internal test AUC = 0.767 vs. external validation AUC = 0.724, P = 0.269) and 0.645 for predicting all cancer (internal test AUC = 0.650 vs. external validation AUC = 0.640, P = 0.848). Integrated model with the mean AUC of 0.803 for predicting csPCa (internal test AUC = 0.804 vs. external validation AUC = 0.801, P = 0.019) and 0.778 for predicting all cancer (internal test AUC = 0.801 vs. external validation AUC = 0.754, P = 0.047).

Conclusions

The radiomics model based on machine learning has the potential to be a non-invasive tool to distinguish cancerous, noncancerous and csPCa in PI-RADS 3 lesions, and has relatively high generalization ability between different date set.

Differential value of diffusion kurtosis imaging and intravoxel incoherent motion in benign and malignant solitary pulmonary lesions

Objective

To investigate the diagnostic value of diffusion kurtosis imaging (DKI) and intravoxel incoherent motion (IVIM) whole-lesion histogram parameters in differentiating benign and malignant solitary pulmonary lesions (SPLs).

Materials and Methods

Patients with SPLs detected by chest CT examination and with further routine MRI, DKI and IVIM-DWI functional sequence scanning data were recruited. According to the pathological results, SPLs were divided into a benign group and a malignant group. Independent samples t tests (normal distribution) or Mann‒Whitney U tests (nonnormal distribution) were used to compare the differences in DKI (Dk, K), IVIM (D, D*, f) and ADC whole-lesion histogram parameters between the benign and malignant SPL groups. The receiver operating characteristic (ROC) curve was used to evaluate the diagnostic efficiency of the histogram parameters and determine the optimal threshold. The area under the curve (AUC) of each histogram parameter was compared by the DeLong method. Spearman rank correlation was used to analyze the correlation between histogram parameters and malignant SPLs.

Results

Most of the histogram parameters for diffusion-related values (Dk, D, ADC) of malignant SPLs were significantly lower than those of benign SPLs, while most of the histogram parameters for the K value of malignant SPLs were significantly higher than those of benign SPLs. DKI (Dk, K), IVIM (D) and ADC were effective in differentiating benign and malignant SPLs and combined with multiple parameters of the whole-lesion histogram for the D value, had the highest diagnostic efficiency, with an AUC of 0.967, a sensitivity of 90.00% and a specificity of 94.03%. Most of the histogram parameters for the Dk, D and ADC values were negatively correlated with malignant SPLs, while most of the histogram parameters for the K value were positively correlated with malignant SPLs.

Conclusions

DKI (Dk, K) and IVIM (D) whole-lesion histogram parameters can noninvasively distinguish benign and malignant SPLs, and the diagnostic performance is better than that of DWI. Moreover, they can provide additional information on SPL microstructure, which has important significance for guiding clinical individualized precision diagnosis and treatment and has potential clinical application value.

Repeatability and test-retest reproducibility of mean apparent diffusion coefficient measurements of focal and diffuse disease in relapsed multiple myeloma at 3T whole body diffusion-weighted MRI (WB-DW-MRI)

OBJECTIVE

To assess the test-retest reproducibility and intra/interobserver agreement of apparent diffusion coefficient (ADC) measurements of myeloma lesions using whole body diffusion-weighted MRI (WB-DW-MRI) at 3T MRI.

METHODS

Following ethical approval, 11 consenting patients with relapsed multiple myeloma were prospectively recruited and underwent baseline WB-DW-MRI. For a single bed position, axial DWI was repeated after a short interval to permit test-retest measurements.Mean ADC measurement was performed by two experienced observers. Intra- and interobserver agreement and test-retest reproducibility were assessed, using coefficient of variation (CV) and interclass correlation coefficient (ICC) measures, for diffuse and focal lesions (small ≤10 mm and large >10 mm).

RESULTS

47 sites of disease were outlined (23 focal, 24 diffuse) in different bed positions (pelvis = 22, thorax = 20, head and neck = 5). For all lesions, there was excellent intraobserver agreement with ICC of 0.99 (0.98-0.99) and COV of 5%. For interobserver agreement, ICC was 0.89 (0.8-0.934) and COV was 17%. There was poor interobserver agreement for diffuse disease (ICC = 0.46) and small lesions (ICC = 0.54).For test-retest reproducibility, excellent ICC (0.916) and COV (14.5%) values for mean ADC measurements were observed. ICCs of test-retest were similar between focal lesions (0.83) and diffuse infiltration (0.80), while ICCs were higher in pelvic (0.95) compared to thoracic (0.81) region and in small (0.96) compared to large (0.8) lesions.

CONCLUSION

ADC measurements of focal lesions in multiple myeloma are repeatable and reproducible, while there is more variation in ADC measurements of diffuse disease in patients with multiple myeloma.

ADVANCES IN KNOWLEDGE

Mean ADC measurements are repeatable and reproducible in focal lesions in multiple myeloma, while the ADC measurements of diffuse disease in multiple myeloma are more subject to variation. The evidence supports the future potential role of ADC measurements as predictive quantitative biomarker in multiple myeloma.

Repeatability and Reproducibility Assessment of the Apparent Diffusion Coefficient in the Prostate: A Trial of the ECOG-ACRIN Research Group (ACRIN 6701)

BACKGROUND

Uncertainty regarding the reproducibility of the apparent diffusion coefficient (ADC) hampers the use of quantitative diffusion-weighted imaging (DWI) in evaluation of the prostate with magnetic resonance imaging MRI. The quantitative imaging biomarkers alliance (QIBA) profile for quantitative DWI claims a within-subject coefficient of variation (wCV) for prostate lesion ADC of 0.17. Improved understanding of ADC reproducibility would aid the use of quantitative diffusion in prostate MRI evaluation.

PURPOSE

Evaluation of the repeatability (same-day) and reproducibility (multi-day) of whole-prostate and focal-lesion ADC assessment in a multi-site setting.

STUDY TYPE

Prospective multi-institutional.

SUBJECTS

Twenty-nine males, ages 53 to 80 (median 63) years, following diagnosis of prostate cancer, 10 with focal lesions.

FIELD STRENGTH/SEQUENCE

3T, single-shot spin-echo diffusion-weighted echo-planar sequence with four b-values.

ASSESSMENT

Sites qualified for the study using an ice-water phantom with known ADC. Readers performed DWI analyses at visit 1 ("V1") and visit 2 ("V2," 2-14 days after V1), where V2 comprised scans before ("V2pre") and after ("V2post") a "coffee-break" interval with subject removal and repositioning. A single reader segmented the whole prostate. Two readers separately placed region-of-interests for focal lesions.

STATISTICAL TESTS

Reproducibility and repeatability coefficients for whole prostate and focal lesions derived from median pixel ADC. We estimated the wCV and 95% confidence interval using a variance stabilizing transformation and assessed interreader reliability of focal lesion ADC using the intraclass correlation coefficient (ICC).

RESULTS

The ADC biases from b0 -b600 and b0 -b800 phantom scans averaged 1.32% and 1.44%, respectively; mean b-value dependence was 0.188%. Repeatability and reproducibility of whole prostate median pixel ADC both yielded wCVs of 0.033 (N = 29). In 10 subjects with an evaluable focal lesion, the individual reader wCVs were 0.148 and 0.074 (repeatability) and 0.137 and 0.078 (reproducibility). All time points demonstrated good to excellent interreader reliability for focal lesion ADC (ICCV1  = 0.89; ICCV2pre  = 0.76; ICCV2post  = 0.94).

DATA CONCLUSION

This study met the QIBA claim for prostate ADC. Test-retest repeatability and multi-day reproducibility were largely equivalent. Interreader reliability for focal lesion ADC was high across time points.

LEVEL OF EVIDENCE

1 TECHNICAL EFFICACY: Stage 2 TOC CATEGORY: Pelvis.

Physically implausible signals as a quantitative quality assessment metric in prostate diffusion-weighted MR imaging

PURPOSE

To provide a quantitative assessment of diffusion-weighted MR images of the prostate through identification of PIDS which clearly represents artifacts in the data. We calculated the percentage and distribution of PIDS in prostate DWI and compare the amount of PIDS between mpMRI images obtained with and without an endorectal coil.

METHODS

This IRB approved retrospective study (from 03/03/2014 to 03/10/2020), included 40 patients scanned with endorectal coil (ERC) and 40 without ER coil (NERC). PIDS contains any voxel where: (1) the diffusion signal increases despite an increase in b-value; and/or (2) apparent diffusion coefficient (ADC) is more than 3.0 μm²/ms (the ADC of pure water at 37 °C and it is physically implausible for any material to have a higher ADC). PIDS for transition zone (TZ) and peripheral zone (PZ) was calculated using an in-house MATLAB program. DWI images were quantitatively inspected for noise, motion, and distortion. T-test was used to compare the difference between PIDS levels in ERC versus NERC and ANOVA to compare the PIDS levels in the anatomic zones. The images were evaluated by a fellowship-trained radiologist in Abdominal Imaging with more than 10 years of experience in reading prostate MRI. This was tested only in prostate in this study.

RESULTS

80 patients (58 ± 8 years old, 80 men) were evaluated. The percentage of voxels exhibiting PIDS was 17.1 ± 8.1% for the ERC cohort and 22.2 ± 15.5% for the NERC cohort. PIDS for NERC versus ERC were not significantly different (p = 0.14). The apex and base showed similar percentages of PIDS in ERC (p = 0.30) and NERC (p = 0.86). The mid (13.8 ± 8.6%) in ERC showed lower values (p = 0.02) of PIDS compared to apex (19.9 ± 11.1%) and base (17.5 ± 8.3%).

CONCLUSION

PIDS maps provide a spatially resolved quantitative quality assessment for prostate DWI. Average PIDS over the entire prostate were similar for the ERC and NERC cohorts, and did not differ significantly across prostate zones. However, for many of the patients, PIDS was focally much higher in specific prostate zones. PIDS assessment can guide Radiologist's evaluation of images and the development of improved DWI sequences.

Whole-tumour apparent diffusion coefficient (ADC) histogram analysis to identify MYCN-amplification in neuroblastomas: preliminary results

OBJECTIVES

To determine the role of apparent diffusion coefficient (ADC) histogram analysis in the identification of MYCN-amplification status in neuroblastomas.

METHODS

We retrospectively evaluated imaging records from 62 patients with neuroblastomas (median age: 15 months (interquartile range (IQR): 7-24 months); 38 females) who underwent magnetic resonance imaging at our institution before the initiation of any therapy or biopsy. Fourteen patients had MYCN-amplified (MYCNA) neuroblastoma. Histogram parameters of ADC maps from the entire tumour was obtained from the baseline images and the normalised images. The Mann-Whitney U test was used to compare the absolute and normalised histogram parameters amongst neuroblastomas with and without MYCN-amplification. Receiver operating characteristic (ROC) curves and area under the curves (AUC) were generated for the statistically significant histogram parameters. Cut-offs obtained from the ROC curves were evaluated on an external validation set (n-15, MYCNA-6, F-7, age 24 months (10-60)). A logistic regression model was trained to predict MYCNA by combining statistically significant histogram parameters and was evaluated on the validation set.

RESULTS

MYCN-amplified neuroblastomas had statistically significant higher maximum ADC and lower minimum ADC than non-amplified neuroblastomas. They also demonstrated higher entropy, variance, energy, and lower uniformity than non-amplified neoplasms (p > 0.05). Energy, entropy, and maximum ADC had AUC of 0.85, 0.79, and 0.82, respectively.

CONCLUSIONS

Whole tumour ADC histogram analysis of neuroblastomas can differentiate between tumours with and without MYCN-amplification. These parameters can help identify areas for targeted biopsies or can be used to predict subtypes of these high-risk tumours before biopsy results are available.

KEY POINTS

• MYCN-amplification significantly affects treatment decisions in neuroblastomas. • MYCN-amplified neuroblastomas had significantly different ADC histogram metrics as compared to tumours without amplification. • ADC histogram metrics can be used to predict MYCN-amplification status based on imaging.

Serial changes in tumour measurements and apparent diffusion coefficients in prostate cancer patients on active surveillance with and without histopathological progression

OBJECTIVE

To analyse serial changes in MRI-derived tumour measurements and apparent diffusion coefficient (ADC) values in prostate cancer (PCa) patients on active surveillance (AS) with and without histopathological disease progression.

METHODS

This study included AS patients with biopsy-proven PCa with a minimum of two consecutive MR examinations and at least one repeat targeted biopsy. Tumour volumes, largest axial two-dimensional (2D) surface areas, and maximum diameters were measured on T2 weighted images (T2WI). ADC values were derived from the whole lesions, 2D areas, and small-volume regions of interest (ROIs) where tumours were most conspicuous. Areas under the ROC curve (AUCs) were calculated for combinations of T2WI and ADC parameters with optimal specificity and sensitivity.

RESULTS

60 patients (30 progressors and 30 non-progressors) were included. In progressors, T2WI-derived tumour volume, 2D surface area, and maximum tumour diameter had a median increase of +99.5%,+55.3%, and +21.7% compared to +29.2%,+8.1%, and +6.9% in non-progressors (p < 0.005 for all). Follow-up whole-volume and small-volume ROIs ADC values were significantly reduced in progressors (-11.7% and -9.5%) compared to non-progressors (-6.1% and -1.6%) (p < 0.05 for both). The combined AUC of a relative increase in maximum tumour diameter by 20% and reduction in small-volume ADC by 10% was 0.67.

CONCLUSION

AS patients show significant differences in tumour measurements and ADC values between those with and without histopathological disease progression.

ADVANCES IN KNOWLEDGE

This paper proposes specific clinical cut-offs for T2WI-derived maximum tumour diameter (+20%) and small-volume ADC (-10%) to predict histopathological PCa progression on AS and supplement subjective serial MRI assessment.

Semi-automatic quantitative analysis of the pelvic bony structures on apparent diffusion coefficient maps based on deep learning: establishment of reference ranges

BACKGROUND

Apparent diffusion coefficient (ADC) maps provide quantitative information on both normal and abnormal tissues. However, it is difficult to distinguish between these tissues unless consistent and precise ADC values can be obtained from normal tissues. For this study we developed a deep learning-based convolutional neural network (CNN) for pelvic bony structure segmentation and established the reference ranges of ADC parameters for normal pelvic bony structures.

METHODS

We retrospectively enrolled 767 prostate cancer (PCa) patients for quantitative ADC analyses of normal pelvic bony structures. A subset of 288 patients who did not receive treatment for PCa (S1) were used to develop a CNN model for the segmentation of 8 pelvic bony structures (lumbar vertebra, sacrococcyx, ilium, acetabulum, femoral head, femoral neck, ischium, and pubis). The proposed CNN was used for the automated segmentation of these pelvic bony structures from a subset of 405 patients who did not receive treatment (S2) and 74 patients who received treatment [radiotherapy (S3) or endocrine therapy (S4)]. The 95% confidence interval (CI) was used to establish reference ranges for the ADC values from the normal pelvic bony structures of S1 and S2.

RESULTS

The Dice scores (Sørensen-Dice coefficient) for the CNN segmentation of the 8 pelvic bones on the ADC maps ranged from 0.90±0.02 (ilium) to 0.95±0.03 (femoral head) in the S1 testing set. In the S2 data set, the Dice scores showed no significant difference among the different scanners (P>0.05), and no significant differences were found among the S2, S3, and S4 data sets. The correlation analysis revealed that the b value and field strength were significantly correlated with ADC values (all P<0.001), while age and treatment were not significant variables (all P>0.05). The ADC reference ranges (95% CI) were as follows: lumbar vertebra, 1.11 (0.90-1.54); sacrococcyx, 0.82 (0.61-1.15); ilium, 0.57 (0.45-0.62); acetabulum, 0.59 (0.40-0.69); femoral head, 0.46 (0.25-0.58); femoral neck, 0.43 (0.25-0.48); ischium, 0.45 (0.26-0.55); and pubis, 0.57 (0.45-0.65).

CONCLUSIONS

This study preliminarily established reference ranges for the ADC values of normal pelvic bony structures. The image acquisition parameters had an influence on the ADC values.

Grading of meningioma tumors based on analyzing tumor volumetric histograms obtained from conventional MRI and apparent diffusion coefficient images

Background

Our purpose was to evaluate the application of volumetric histogram parameters obtained from conventional MRI and apparent diffusion coefficient (ADC) images for grading the meningioma tumors.

Results

Tumor volumetric histograms of preoperative MRI images from 45 patients with the diagnosis of meningioma at different grades were analyzed to find the histogram parameters. Kruskal-Wallis statistical test was used for comparison between the parameters obtained from different grades. Multi-parametric regression analysis was used to find the model and parameters with high predictive value for the classification of meningioma. Mode; standard deviation on post-contrast T1WI, T2-FLAIR, and ADC images; kurtosis on post-contrast T1WI and T2-FLAIR images; mean and several percentile values on ADC; and post-contrast T1WI images showed significant differences among different tumor grades (P < 0.05). The multi-parametric linear regression showed that the ADC histogram parameters model had a higher predictive value, with cutoff values of 0.212 (sensitivity = 79.6%, specificity = 84.3%) and 0.180 (sensitivity = 70.9%, specificity = 80.8%) for differentiating the grade I from II, and grade II from III, respectively.

Conclusions

The multi-parametric model of volumetric histogram parameters in some of the conventional MRI series (i.e., post-contrast T1WI and T2-FLAIR images) along with the ADC images are appropriate for predicting the meningioma tumors’ grade.

Utility of machine learning of apparent diffusion coefficient (ADC) and T2-weighted (T2W) radiomic features in PI-RADS version 2.1 category 3 lesions to predict prostate cancer diagnosis

PURPOSE

To evaluate if machine learning (ML) of radiomic features extracted from apparent diffusion coefficient (ADC) and T2-weighted (T2W) MRI can predict prostate cancer (PCa) diagnosis in Prostate Imaging-Reporting and Data System (PI-RADS) version 2.1 category 3 lesions.

METHODS

This multi-institutional review board-approved retrospective case-control study evaluated 158 men with 160 PI-RADS category 3 lesions (79 peripheral zone, 81 transition zone) diagnosed at 3-Tesla MRI with histopathology diagnosis by MRI-TRUS-guided targeted biopsy. A blinded radiologist confirmed PI-RADS v2.1 score and segmented lesions on axial T2W and ADC images using 3D Slicer, extracting radiomic features with an open-source software (Pyradiomics). Diagnostic accuracy for (1) any PCa and (2) clinically significant (CS; International Society of Urogenital Pathology Grade Group ≥ 2) PCa was assessed using XGBoost with tenfold cross -validation.

RESULTS

From 160 PI-RADS 3 lesions, there were 50.0% (80/160) PCa, including 36.3% (29/80) CS-PCa (63.8% [51/80] ISUP 1, 23.8% [19/80] ISUP 2, 8.8% [7/80] ISUP 3, 3.8% [3/80] ISUP 4). The remaining 50.0% (80/160) lesions were benign. ML of all radiomic features from T2W and ADC achieved area under receiver operating characteristic curve (AUC) for diagnosis of (1) CS-PCa 0.547 (95% Confidence Intervals 0.510-0.584) for T2W and 0.684 (CI 0.652-0.715) for ADC and (2) any PCa 0.608 (CI 0.579-0.636) for T2W and 0.642 (CI 0.614-0.0.670) for ADC.

CONCLUSION

Our results indicate ML of radiomic features extracted from T2W and ADC achieved at best moderate accuracy for determining which PI-RADS category 3 lesions represent PCa.

The Histogram Analysis of Intravoxel Incoherent Motion-Kurtosis Model in the Diagnosis and Grading of Prostate Cancer-A Preliminary Study

Objectives

This study was conducted in order to explore the value of histogram analysis of the intravoxel incoherent motion-kurtosis (IVIM-kurtosis) model in the diagnosis and grading of prostate cancer (PCa), compared with monoexponential model (MEM).

Materials and Methods

Thirty patients were included in this study. Single-shot echo-planar imaging (SS-EPI) diffusion-weighted images (b-values of 0, 20, 50, 100, 200, 500, 1,000, 1,500, 2,000 s/mm²) were acquired. The pathologies were confirmed by in-bore MR-guided biopsy. The postprocessing and measurements were processed using the software tool Matlab R2015b for the IVIM-kurtosis model and MEM. Regions of interest (ROIs) were drawn manually. Mean values of D, D*, f, K, ADC, and their histogram parameters were acquired. The values of these parameters in PCa and benign prostatic hyperplasia (BPH)/prostatitis were compared. Receiver operating characteristic (ROC) curves were used to investigate the diagnostic efficiency. The Spearman test was used to evaluate the correlation of these parameters and Gleason scores (GS) of PCa.

Results

For the IVIM-kurtosis model, D (mean, 10th, 25th, 50th, 75th, 90th), D* (90th), and f (10th) were significantly lower in PCa than in BPH/prostatitis, while D (skewness), D* (kurtosis), and K (mean, 75th, 90th) were significantly higher in PCa than in BPH/prostatitis. For MEM, ADC (mean, 10th, 25th, 50th, 75th, 90th) was significantly lower in PCa than in BPH/prostatitis. The area under the ROC curve (AUC) of the IVIM-kurtosis model was higher than MEM, without significant differences (z = 1.761, P = 0.0783). D (mean, 50th, 75th, 90th), D* (mean, 10th, 25th, 50th, 75th), and f (skewness, kurtosis) correlated negatively with GS, while D (kurtosis), D* (skewness, kurtosis), f (mean, 75th, 90th), and K (mean, 75th, 90th) correlated positively with GS. The histogram parameters of ADC did not show correlations with GS.

Conclusion

The IVIM-kurtosis model has potential value in the differential diagnosis of PCa and BPH/prostatitis. IVIM-kurtosis histogram analysis may provide more information in the grading of PCa than MEM.

Whole tumor volumetric ADC analysis: relationships with histopathological differentiation of hepatocellular carcinoma

PURPOSE

The aim of this study was to investigate the relationships between values obtained from whole tumor volumetric apparent diffusion coefficient (ADC) measurements and histopathological grade in patients with hepatocellular carcinoma (HCC).

METHODS

Fifty-one naïve patients with HCC were included in the study. The tumors were classified according to the Edmondson-Steiner grade and separated as well-differentiated and non-well-differentiated (moderately and poorly differentiated). The ADC parameters of groups were compared by applying Mann-Whitney U test. The correlation between tumors' histopathological stage and whole tumor ADC parameters was investigated using Spearman's Rank Correlation Coefficient. The receiver operating characteristic curve analysis (ROC) was applied to calculate the area under curve (AUC) with intersection point of ADC parameters and curve.

RESULTS

Mean and percentile ADC values of well-differentiated tumors were significantly higher than those of non-well-differentiated tumors (p < 0.05). The strongest correlation between histopathological grade and ADC parameters was 75th percentile ADC (r = - 0.501), 50th percentile ADC (r = - 0.476) and mean ADC (r = - 0.465). Mean, 75th and 50th percentile ADC values used for the distinction of groups gave the highest AUC at ROC analysis (0.781, 0.781, 0.767, respectively). When threshold values of mean, 75th and 50th percentile ADC values were applied (1516 mm²/s, 1194 mm²/s, and 1035 mm²/s) sensitivity was calculated as 0.73, 0.91, 0.83, respectively, and specificity was calculated as 0.82, 0.61, and 0.68, respectively.

CONCLUSIONS

A correlation between whole tumor volumetric ADC values and HCCs' histopathological grade was detected in this study. 75th percentile, 50th percentile and mean ADC values are determined as highly sensitive and specific tests when the threshold values are applied for distinguishing between well-differentiated tumors and moderately/poorly differentiated tumors. When all these findings are evaluated together, HCCs' volumetric ADC values might be a useful noninvasive predictive parameters for histopathological grade in patients with HCC.

Multiparametric MRI for assessment of early response to neoadjuvant sunitinib in renal cell carcinoma

PURPOSE

To detect early response to sunitinib treatment in metastatic clear cell renal cancer (mRCC) using multiparametric MRI.

METHOD

Participants with mRCC undergoing pre-surgical sunitinib therapy in the prospective NeoSun clinical trial (EudraCtNo: 2005-004502-82) were imaged before starting treatment, and after 12 days of sunitinib therapy using morphological MRI sequences, advanced diffusion-weighted imaging, measurements of R2* (related to hypoxia) and dynamic contrast-enhanced imaging. Following nephrectomy, participants continued treatment and were followed-up with contrast-enhanced CT. Changes in imaging parameters before and after sunitinib were assessed with the non-parametric Wilcoxon signed-rank test and the log-rank test was used to assess effects on survival.

RESULTS

12 participants fulfilled the inclusion criteria. After 12 days, the solid and necrotic tumor volumes decreased by 28% and 17%, respectively (p = 0.04). However, tumor-volume reduction did not correlate with progression-free or overall survival (PFS/OS). Sunitinib therapy resulted in a reduction in median solid tumor diffusivity D from 1298x10-6 to 1200x10-6mm2/s (p = 0.03); a larger decrease was associated with a better RECIST response (p = 0.02) and longer PFS (p = 0.03) on the log-rank test. An increase in R2* from 19 to 28s-1 (p = 0.001) was observed, paralleled by a decrease in Ktrans from 0.415 to 0.305min-1 (p = 0.01) and a decrease in perfusion fraction from 0.34 to 0.19 (p<0.001).

CONCLUSIONS

Physiological imaging confirmed efficacy of the anti-angiogenic agent 12 days after initiating therapy and demonstrated response to treatment. The change in diffusivity shortly after starting pre-surgical sunitinib correlated to PFS in mRCC undergoing nephrectomy, however, no parameter predicted OS.

TRIAL REGISTRATION

EudraCtNo: 2005-004502-82.

Effects of different breathing techniques on the IVIM-derived quantitative parameters of the normal pancreas

PURPOSE

To prospectively compare the differences in intravoxel incoherent motion (IVIM)-derived quantitative parameters in different anatomic locations of the normal pancreas with different breathing techniques in a healthy population.

METHOD

Twenty-six volunteers successfully underwent pancreas axial IVIM imaging with a 3.0-T MR system using 11 b-values (from 0 to 1000 sec/mm²) with three different breathing techniques: free breath (FB), liver dome scout (LDS), and phase scout (PS). The IVIM-derived quantitative parameters in three anatomic locations (head, body, and tail of the pancreas) were calculated. The intra-, inter-, and short-term consistency of IVIM-derived quantitative parameters were assessed by comparing 95% confidence interval (CI) of limits of agreement (LOA) of difference between measurements and clinical maximum allowed difference using the Bland-Altman method. The Kruskal-Wallis test was used to compare pancreatic IVIM-derived parameters.

RESULTS

In Bland-Altman graph, the maximum values of the 95% CIs of LOAs of D_slow, D_fast, and f were (0.123 ± 0.022) × 10^-3 mm²/sec, (22.093 ± 4.997) × 10^-3 mm²/sec, and (3.942 ± 0.621)%, and the consistency of D_slow and f was good and that of D_fast was poor overall. The D_slow, D_fast, and f values of normal pancreas were (1.056 ± 0.121) × 10^-3 mm²/sec, (55.755 ± 13.011) × 10^-3 mm²/sec, and (26.036 ± 2.361)%, respectively, and there aren't any breathing technique (P > 0.05) or location (P > 0.05) dependent differences.

CONCLUSIONS

Our study shows that IVIM-derived quantitative parameters of the pancreas may not be affected by breathing techniques and anatomic locations. The f and D_slow values have good repeated measurement consistency under different breathing techniques.

Reproducibility of magnetic resonance fingerprinting-based T1 mapping of the healthy prostate at 1.5 and 3.0 T: A proof-of-concept study

Facilitating clinical translation of quantitative imaging techniques has been suggested as means of improving interobserver agreement and diagnostic accuracy of multiparametric magnetic resonance imaging (mpMRI) of the prostate. One such technique, magnetic resonance fingerprinting (MRF), has significant competitive advantages over conventional mapping techniques in terms of its multi-site reproducibility, short scanning time and inherent robustness to motion. It has also been shown to improve the detection of clinically significant prostate cancer when added to standard mpMRI sequences, however, the existing studies have all been conducted on 3.0 T MRI systems, limiting the technique's use on 1.5 T MRI scanners that are still more widely used for prostate imaging across the globe. The aim of this proof-of-concept study was, therefore, to evaluate the cross-system reproducibility of prostate MRF T1 in healthy volunteers (HVs) using 1.5 and 3.0 T MRI systems. The initial validation of MRF T1 against gold standard inversion recovery fast spin echo (IR-FSE) T1 in the ISMRM/NIST MRI system revealed a strong linear correlation between phantom-derived MRF and IR-FSE T1 values was observed at both field strengths (R2 = 0.998 at 1.5T and R2 = 0.993 at 3T; p = < 0.0001 for both). In young HVs, inter-scanner CVs demonstrated marginal differences across all tissues with the highest difference of 3% observed in fat (2% at 1.5T vs 5% at 3T). At both field strengths, MRF T1 could confidently differentiate prostate peripheral zone from transition zone, which highlights the high quantitative potential of the technique given the known difficulty of tissue differentiation in this age group. The high cross-system reproducibility of MRF T1 relaxometry of the healthy prostate observed in this preliminary study, therefore, supports the technique's prospective clinical validation as part of larger trials employing 1.5 T MRI systems, which are still widely used clinically for routine mpMRI of the prostate.

Image Contrast, Image Pre-Processing, and T1 Mapping Affect MRI Radiomic Feature Repeatability in Patients with Colorectal Cancer Liver Metastases

Imaging biomarkers require technical, biological, and clinical validation to be translated into robust tools in research or clinical settings. This study contributes to the technical validation of radiomic features from magnetic resonance imaging (MRI) by evaluating the repeatability of features from four MR sequences: pre-contrast T1- and T2-weighted images, pre-contrast quantitative T1 maps (qT1), and contrast-enhanced T1-weighted images. Fifty-one patients with colorectal cancer liver metastases were scanned twice, up to 7 days apart. Repeatability was quantified using the intraclass correlation coefficient (ICC) and repeatability coefficient (RC), and the impact of non-Gaussian feature distributions and image normalisation was evaluated. Most radiomic features had non-Gaussian distributions, but Box-Cox transformations enabled ICCs and RCs to be calculated appropriately for an average of 97% of features across sequences. ICCs ranged from 0.30 to 0.99, with volume and other shape features tending to be most repeatable; volume ICC > 0.98 for all sequences. 19% of features from non-normalised images exhibited significantly different ICCs in pair-wise sequence comparisons. Normalisation tended to increase ICCs for pre-contrast T1- and T2-weighted images, and decrease ICCs for qT1 maps. RCs tended to vary more between sequences than ICCs, showing that evaluations of feature performance depend on the chosen metric. This work suggests that feature-specific repeatability, from specific combinations of MR sequence and pre-processing steps, should be evaluated to select robust radiomic features as biomarkers in specific studies. In addition, as different repeatability metrics can provide different insights into a specific feature, consideration of the appropriate metric should be taken in a study-specific context.

Accelerating Prostate Diffusion-weighted MRI Using a Guided Denoising Convolutional Neural Network: Retrospective Feasibility Study

PURPOSE

To investigate the feasibility of accelerating prostate diffusion-weighted imaging (DWI) by reducing the number of acquired averages and denoising the resulting image using a proposed guided denoising convolutional neural network (DnCNN).

MATERIALS AND METHODS

Raw data from the prostate DWI scans were retrospectively gathered between July 2018 and July 2019 from six single-vendor MRI scanners. There were 103 datasets used for training (median age, 64 years; interquartile range [IQR], 11), 15 for validation (median age, 68 years; IQR, 12), and 37 for testing (median age, 64 years; IQR, 12). High b-value diffusion-weighted (hb DW) data were reconstructed into noisy images using two averages and reference images using all 16 averages. A conventional DnCNN was modified into a guided DnCNN, which uses the low b-value DW image as a guidance input. Quantitative and qualitative reader evaluations were performed on the denoised hb DW images. A cumulative link mixed regression model was used to compare the readers' scores. The agreement between the apparent diffusion coefficient (ADC) maps (denoised vs reference) was analyzed using Bland-Altman analysis.

RESULTS

Compared with the original DnCNN, the guided DnCNN produced denoised hb DW images with higher peak signal-to-noise ratio (32.79 ± 3.64 [standard deviation] vs 33.74 ± 3.64), higher structural similarity index (0.92 ± 0.05 vs 0.93 ± 0.04), and lower normalized mean square error (3.9% ± 10 vs 1.6% ± 1.5) (P < .001 for all). Compared with the reference images, the denoised images received higher image quality scores from the readers (P < .0001). The ADC values based on the denoised hb DW images were in good agreement with the reference ADC values (mean ADC difference ranged from -0.04 to 0.02 × 10^-3 mm²/sec).

CONCLUSION

Accelerating prostate DWI by reducing the number of acquired averages and denoising the resulting image using the proposed guided DnCNN is technically feasible. Supplemental material is available for this article. © RSNA, 2020.

Prostate MRI radiomics: A systematic review and radiomic quality score assessment

BACKGROUND

Radiomics have the potential to further increase the value of MRI in prostate cancer management. However, implementation in clinical practice is still far and concerns have been raised regarding the methodological quality of radiomic studies. Therefore, we aimed to systematically review the literature to assess the quality of prostate MRI radiomic studies using the radiomics quality score (RQS).

METHODS

Multiple medical literature archives (PubMed, Web of Science and EMBASE) were searched to retrieve original investigations focused on prostate MRI radiomic approaches up to the end of June 2019. Three researchers independently assessed each paper using the RQS. Data from the most experienced researcher were used for descriptive analysis. Inter-rater reproducibility was assessed using the intraclass correlation coefficient (ICC) on the total RQS score.

RESULTS

73 studies were included in the analysis. Overall, the average RQS total score was 7.93 ± 5.13 on a maximum of 36 points, with a final average percentage of 23 ± 13%. Among the most critical items, the lack of feature robustness testing strategies and external validation datasets. The ICC resulted poor to moderate, with an average value of 0.57 and 95% Confidence Intervals between 0.44 and 0.69.

CONCLUSIONS

Current studies on prostate MRI radiomics still lack the quality required to allow their introduction in clinical practice.

Quantitative Prostate MRI

Prostate MRI is reported in clinical practice using the Prostate Imaging and Data Reporting System (PI-RADS). PI-RADS aims to standardize, as much as possible, the acquisition, interpretation, reporting, and ultimately the performance of prostate MRI. PI-RADS relies upon mainly subjective analysis of MR imaging findings, with very few incorporated quantitative features. The shortcomings of PI-RADS are mainly: low-to-moderate interobserver agreement and modest accuracy for detection of clinically significant tumors in the transition zone. The use of a more quantitative analysis of prostate MR imaging findings is therefore of interest. Quantitative MR imaging features including: tumor size and volume, tumor length of capsular contact, tumor apparent diffusion coefficient (ADC) metrics, tumor T₁ and T₂ relaxation times, tumor shape, and texture analyses have all shown value for improving characterization of observations detected on prostate MRI and for differentiating between tumors by their pathological grade and stage. Quantitative analysis may therefore improve diagnostic accuracy for detection of cancer and could be a noninvasive means to predict patient prognosis and guide management. Since quantitative analysis of prostate MRI is less dependent on an individual users' assessment, it could also improve interobserver agreement. Semi- and fully automated analysis of quantitative (radiomic) MRI features using artificial neural networks represent the next step in quantitative prostate MRI and are now being actively studied. Validation, through high-quality multicenter studies assessing diagnostic accuracy for clinically significant prostate cancer detection, in the domain of quantitative prostate MRI is needed. This article reviews advances in quantitative prostate MRI, highlighting the strengths and limitations of existing and emerging techniques, as well as discussing opportunities and challenges for evaluation of prostate MRI in clinical practice when using quantitative assessment. LEVEL OF EVIDENCE: 5 TECHNICAL EFFICACY: Stage 2.

A Hybrid End-to-End Approach Integrating Conditional Random Fields into CNNs for Prostate Cancer Detection on MRI

Prostate Cancer (PCa) is the most common oncological disease in Western men. Even though a growing effort has been carried out by the scientific community in recent years, accurate and reliable automated PCa detection methods on multiparametric Magnetic Resonance Imaging (mpMRI) are still a compelling issue. In this work, a Deep Neural Network architecture is developed for the task of classifying clinically significant PCa on non-contrast-enhanced MR images. In particular, we propose the use of Conditional Random Fields as a Recurrent Neural Network (CRF-RNN) to enhance the classification performance of XmasNet, a Convolutional Neural Network (CNN) architecture specifically tailored to the PROSTATEx17 Challenge. The devised approach builds a hybrid end-to-end trainable network, CRF-XmasNet, composed of an initial CNN component performing feature extraction and a CRF-based probabilistic graphical model component for structured prediction, without the need for two separate training procedures. Experimental results show the suitability of this method in terms of classification accuracy and training time, even though the high-variability of the observed results must be reduced before transferring the resulting architecture to a clinical environment. Interestingly, the use of CRFs as a separate postprocessing method achieves significantly lower performance with respect to the proposed hybrid end-to-end approach. The proposed hybrid end-to-end CRF-RNN approach yields excellent peak performance for all the CNN architectures taken into account, but it shows a high-variability, thus requiring future investigation on the integration of CRFs into a CNN.

Molecular imaging of the prostate: Comparing total sodium concentration quantification in prostate cancer and normal tissue using dedicated 13 C and 23 Na endorectal coils

BACKGROUND

There has been recent interest in nonproton MRI including hyperpolarized carbon-13 (13 C) imaging. Prostate cancer has been shown to have a higher tissue sodium concentration (TSC) than normal tissue. Sodium (23 Na) and 13 C nuclei have a frequency difference of only 1.66 MHz at 3T, potentially enabling 23 Na imaging with a 13 C-tuned coil and maximizing the metabolic information obtained from a single study.

PURPOSE

To compare TSC measurements from a 13 C-tuned endorectal coil to those quantified with a dedicated 23 Na-tuned coil.

STUDY TYPE

Prospective.

POPULATION

Eight patients with biopsy-proven, intermediate/high risk prostate cancer imaged prior to prostatectomy.

SEQUENCE

3T MRI with separate dual-tuned 1 H/23 Na and 1 H/13 C endorectal receive coils to quantify TSC.

ASSESSMENT

Regions-of-interest for TSC quantification were defined for normal peripheral zone (PZ), normal transition zone (TZ), and tumor, with reference to histopathology maps.

STATISTICAL TESTS

Two-sided Wilcoxon rank sum with additional measures of correlation, coefficient of variation, and Bland-Altman plots to assess for between-test differences.

RESULTS

Mean TSC for normal PZ and TZ were 39.2 and 33.9 mM, respectively, with the 23 Na coil and 40.1 and 36.3 mM, respectively, with the 13 C coil (P = 0.22 and P = 0.11 for the intercoil comparison, respectively). For tumor tissue, there was no statistical difference between the overall mean tumor TSC measured with the 23 Na coil (41.8 mM) and with the 13 C coil (46.6 mM; P = 0.38). Bland-Altman plots showed good repeatability for tumor TSC measurements between coils, with a reproducibility coefficient of 9 mM; the coefficient of variation between the coils was 12%. The Pearson correlation coefficient for TSC between coils for all measurements was r = 0.71 (r2 = 0.51), indicating a strong positive linear relationship. The mean TSC within PZ tumors was significantly higher compared with normal PZ for both the 23 Na coil (45.4 mM; P = 0.02) and the 13 C coil (49.4 mM; P = 0.002).

DATA CONCLUSION

We demonstrated the feasibility of using a carbon-tuned coil to quantify TSC, enabling dual metabolic information from a single coil. This approach could make the acquisition of both 23 Na-MRI and 13 C-MRI feasible in a single clinical imaging session.

LEVEL OF EVIDENCE

2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2020;51:90-97.

Diffusion kurtosis MRI as a predictive biomarker of response to neoadjuvant chemotherapy in high grade serous ovarian cancer

This study assessed the feasibility of using diffusion kurtosis imaging (DKI) as a measure of tissue heterogeneity and proliferation to predict the response of high grade serous ovarian cancer (HGSOC) to neoadjuvant chemotherapy (NACT). Seventeen patients with HGSOC were imaged at 3 T and had biopsy samples taken prior to any treatment. The patients were divided into two groups: responders and non-responders based on Response Evaluation Criteria In Solid Tumours (RECIST) criteria. The following imaging metrics were calculated: apparent diffusion coefficient (ADC), apparent diffusion (Dapp) and apparent kurtosis (Kapp). Tumour cellularity and proliferation were quantified using histology and Ki-67 immunohistochemistry. Mean Kapp before therapy was higher in responders compared to non-responders: 0.69 ± 0.13 versus 0.51 ± 0.11 respectively, P = 0.02. Tumour cellularity correlated positively with Kapp (rho = 0.50, P = 0.04) and negatively with both ADC (rho = -0.72, P = 0.001) and Dapp (rho = -0.80, P < 0.001). Ki-67 expression correlated with Kapp (rho = 0.53, P = 0.03) but not with ADC or Dapp. In conclusion, Kapp was found to be a potential predictive biomarker of NACT response in HGSOC, which suggests that DKI is a promising clinical tool for use oncology and radiology that should be evaluated further in future larger studies.

Prostate Imaging Reporting and Data System 3 Category Cases at Multiparametric Magnetic Resonance for Prostate Cancer: A Systematic Review and Meta-analysis

CONTEXT

The Prostate Imaging Reporting and Data System (PI-RADS) 3 score represents a "grey zone" that need to be further investigated to solve the issue of whether to biopsy these equivocal cases or not.

OBJECTIVE

To critically analyze the current evidence on PI-RADS 3 cases. We evaluated the prevalence of PI-RADS 3 cases in the literature and detection rate of prostate cancer (PC) and clinically significant PC (csPC) at biopsy with regard to factors determining these rates.

EVIDENCE ACQUISITION

We searched in the Medline and Cochrane Library database from the literature from January 2009 to January 2019, following the Preferred Reporting Items for Systematic Review and Meta-analyses (PRISMA) guidelines.

EVIDENCE SYNTHESIS

A total of 28 studies were included in our analysis (total number of PI-RADS 3 cases: 1759, range 20-187). The prevalence of PI-RADS 3 cases reported in available studies was 17.3% (range 6.4-45.7%). The PC detection rate was 36% (95% confidence interval [CI] 33.8-37.4; range 10.3-55.8%), whereas that of csPC was 18.5% (95% CI 16.6-20.3; range 3.4-46.5%). Detection rates of PC and csPC were found to be similar in men who underwent a target biopsy versus those with a systematic biopsy (23.5% vs 23.9% and 11.4% vs 12.3%, respectively) and lower than the rates achieved with the combined strategy (36.9% and 19.6%, respectively). A prostate-specific antigen density (PSAD) of ≥0.15ng/ml/ml may represent an index to decide whether to submit a PI-RADS 3 case to biopsy.

CONCLUSIONS

In most investigations, PI-RADS 3 cases were not evaluated separately. A PI-RADS 3 lesion remains an equivocal lesion. Evaluation of clinical predictive factors in terms of csPC risk is a main aspect of helping clinicians in the biopsy decision process.

PATIENT SUMMARY

Management of Prostate Imaging Reporting and Data System 3 cases remains an unmet need, and the detection rate of clinically significant prostate cancer (csPC) among this population varies widely. Performing a combined target plus a systematic biopsy yields the highest detection of csPC. A prostate-specific antigen density of lower than 0.15ng/ml/ml may select patients for a follow-up strategy.

Changes in ADC and T2-weighted MRI-derived radiomic features in patients treated with focal salvage HDR prostate brachytherapy for local recurrence after previous external-beam radiotherapy

PURPOSE

To explore the changes in T2-weighted (T2w) and apparent diffusion coefficient (ADC) magnetic resonance imaging -derived radiomic features of the gross tumor volume (GTV) from focal salvage high-dose-rate prostate brachytherapy (HDRB) and to correlate with clinical parameters.

MATERIALS AND METHODS

Eligible patients included those with biopsy-confirmed local recurrence that correlated with MRI (T2w, ADC). Patients received 27 Gy in 2 fractions separated by 1 week to a quadrant consisting of the GTV. The MRI was repeated 1 year after HDRB. GTVs, planning target volumes, and normal prostate tissue control volumes were identified on the pre- and post-HDRB MRIs. Radiomic features from each GTV were extracted, and principle component analysis identified features with the highest variance.

RESULTS

Pre- and post-HDRB MRIs were obtained from 14 trial patients. Principle component analysis showed that 18 and 17 features contributed to 93% and 86% of the variance observed in the T2w and ADC data, respectively. Sixteen T2w features and 1 ADC GTV feature were different from the control volumes in the pre-HDRB images (p < 0.05). Ten T2w and 7 ADC GTV post-HDRB features were different from those of pre-HDRB (p < 0.05).

CONCLUSIONS

Exploratory analysis reveals several radiomic features in the T2w and ADC image GTVs that distinguish the GTV from healthy prostate tissue and change significantly after salvage HDRB.

[Diffusion-weighted imaging-diagnostic supplement or alternative to contrast agents in early detection of malignancies?]

Medical research in the field of oncologic imaging diagnostics using magnetic resonance imaging increasingly includes diffusion-weighted imaging (DWI) sequences. The DWI sequences allow insights into different microstructural diffusion properties of water molecules in tissues depending on the sequence modification used and enable visual and quantitative analysis of the acquired imaging data. In DWI, the application of intravenous gadolinium-containing contrast agents is unnecessary and only the mobility of naturally occurring water molecules in tissues is quantified. These characteristics predispose DWI as a potential candidate for emerging as an independent diagnostic tool in selected cases and specific points in question. Current clinical diagnostic studies and the ongoing technical developments, including the increasing influence of artificial intelligence in radiology, support the growing importance of DWI. Especially with respect to selective approaches for early detection of malignancies, DWI could make an essential contribution as an eligible diagnostic tool; however, prior to discussing a broader clinical implementation, challenges regarding reliable data quality, standardization and quality assurance must be overcome.