Prostate cancer risk assessment and avoidance of prostate biopsies using fully automatic deep learning in prostate MRI: comparison to PI-RADS and integration with clinical data in nomograms.

OBJECTIVES: Risk calculators (RCs) improve patient selection for prostate biopsy with clinical/demographic information, recently with prostate MRI using the prostate imaging reporting and data system (PI-RADS). Fully-automated deep learning (DL) analyzes MRI data independently, and has been shown to be on par with clinical radiologists, but has yet to be incorporated into RCs. The goal of this study is to re-assess the diagnostic quality of RCs, the impact of replacing PI-RADS with DL predictions, and potential performance gains by adding DL besides PI-RADS. MATERIAL AND METHODS: One thousand six hundred twenty-seven consecutive examinations from 2014 to 2021 were included in this retrospective single-center study, including 517 exams withheld for RC testing. Board-certified radiologists assessed PI-RADS during clinical routine, then systematic and MRI/Ultrasound-fusion biopsies provided histopathological ground truth for significant prostate cancer (sPC). nnUNet-based DL ensembles were trained on biparametric MRI predicting the presence of sPC lesions (UNet-probability) and a PI-RADS-analogous five-point scale (UNet-Likert). Previously published RCs were validated as is; with PI-RADS substituted by UNet-Likert (UNet-Likert-substituted RC); and with both UNet-probability and PI-RADS (UNet-probability-extended RC). Together with a newly fitted RC using clinical data, PI-RADS and UNet-probability, existing RCs were compared by receiver-operating characteristics, calibration, and decision-curve analysis. RESULTS: Diagnostic performance remained stable for UNet-Likert-substituted RCs. DL contained complementary diagnostic information to PI-RADS. The newly-fitted RC spared 49% [252/517] of biopsies while maintaining the negative predictive value (94%), compared to PI-RADS ≥ 4 cut-off which spared 37% [190/517] (p < 0.001). CONCLUSIONS: Incorporating DL as an independent diagnostic marker for RCs can improve patient stratification before biopsy, as there is complementary information in DL features and clinical PI-RADS assessment. CLINICAL RELEVANCE STATEMENT: For patients with positive prostate screening results, a comprehensive diagnostic workup, including prostate MRI, DL analysis, and individual classification using nomograms can identify patients with minimal prostate cancer risk, as they benefit less from the more invasive biopsy procedure. KEY POINTS: The current MRI-based nomograms result in many negative prostate biopsies. The addition of DL to nomograms with clinical data and PI-RADS improves patient stratification before biopsy. Fully automatic DL can be substituted for PI-RADS without sacrificing the quality of nomogram predictions. Prostate nomograms show cancer detection ability comparable to previous validation studies while being suitable for the addition of DL analysis.

A European Society of Oncologic Imaging (ESOI) survey on the radiological assessment of response to oncologic treatments in clinical practice.

OBJECTIVES: To present the results of a survey on the assessment of treatment response with imaging in oncologic patient, in routine clinical practice. The survey was promoted by the European Society of Oncologic Imaging to gather information for the development of reporting models and recommendations. METHODS: The survey was launched on the European Society of Oncologic Imaging website and was available for 3 weeks. It consisted of 5 sections, including 24 questions related to the following topics: demographic and professional information, methods for lesion measurement, how to deal with diminutive lesions, how to report baseline and follow-up examinations, which previous studies should be used for comparison, and role of RECIST 1.1 criteria in the daily clinical practice. RESULTS: A total of 286 responses were received. Most responders followed the RECIST 1.1 recommendations for the measurement of target lesions and lymph nodes and for the assessment of tumor response. To assess response, 48.6% used previous and/or best response study in addition to baseline, 25.2% included the evaluation of all main time points, and 35% used as the reference only the previous study. A considerable number of responders used RECIST 1.1 criteria in daily clinical practice (41.6%) or thought that they should be always applied (60.8%). CONCLUSION: Since standardized criteria are mainly a prerogative of clinical trials, in daily routine, reporting strategies are left to radiologists and oncologists, which may issue local and diversified recommendations. The survey emphasizes the need for more generally applicable rules for response assessment in clinical practice. CRITICAL RELEVANCE STATEMENT: Compared to clinical trials which use specific criteria to evaluate response to oncological treatments, the free narrative report usually adopted in daily clinical practice may lack clarity and useful information, and therefore, more structured approaches are needed. KEY POINTS: · Most radiologists consider standardized reporting strategies essential for an objective assessment of tumor response in clinical practice. · Radiologists increasingly rely on RECIST 1.1 in their daily clinical practice. · Treatment response evaluation should require a complete analysis of all imaging time points and not only of the last.

Deep Learning-based Assessment of Oncologic Outcomes from Natural Language Processing of Structured Radiology Reports.

Purpose: To train a deep natural language processing (NLP) model, using data mined structured oncology reports (SOR), for rapid tumor response category (TRC) classification from free-text oncology reports (FTOR) and to compare its performance with human readers and conventional NLP algorithms. Materials and Methods: In this retrospective study, databases of three independent radiology departments were queried for SOR and FTOR dated from March 2018 to August 2021. An automated data mining and curation pipeline was developed to extract Response Evaluation Criteria in Solid Tumors-related TRCs for SOR for ground truth definition. The deep NLP bidirectional encoder representations from transformers (BERT) model and three feature-rich algorithms were trained on SOR to predict TRCs in FTOR. Models' F1 scores were compared against scores of radiologists, medical students, and radiology technologist students. Lexical and semantic analyses were conducted to investigate human and model performance on FTOR. Results: Oncologic findings and TRCs were accurately mined from 9653 of 12 833 (75.2%) queried SOR, yielding oncology reports from 10 455 patients (mean age, 60 years ± 14 [SD]; 5303 women) who met inclusion criteria. On 802 FTOR in the test set, BERT achieved better TRC classification results (F1, 0.70; 95% CI: 0.68, 0.73) than the best-performing reference linear support vector classifier (F1, 0.63; 95% CI: 0.61, 0.66) and technologist students (F1, 0.65; 95% CI: 0.63, 0.67), had similar performance to medical students (F1, 0.73; 95% CI: 0.72, 0.75), but was inferior to radiologists (F1, 0.79; 95% CI: 0.78, 0.81). Lexical complexity and semantic ambiguities in FTOR influenced human and model performance, revealing maximum F1 score drops of -0.17 and -0.19, respectively. Conclusion: The developed deep NLP model reached the performance level of medical students but not radiologists in curating oncologic outcomes from radiology FTOR.Keywords: Neural Networks, Computer Applications-Detection/Diagnosis, Oncology, Research Design, Staging, Tumor Response, Comparative Studies, Decision Analysis, Experimental Investigations, Observer Performance, Outcomes Analysis Supplemental material is available for this article. © RSNA, 2022.

Dose-efficient assessment of trabecular microstructure using ultra-high-resolution photon-counting CT.

Photon-counting (PC) detectors for clinical computed tomography (CT) may offer improved imaging capabilities compared to conventional energy-integrating (EI) detectors, e.g. superior spatial resolution and detective efficiency. We here investigate if PCCT can reduce the administered dose in examinations aimed at quantifying trabecular bone microstructure. Five human vertebral bodies were scanned three times in an abdomen phantom (QRM, Germany) using an experimental dual-source CT (Somatom CounT, Siemens Healthineers, Germany) housing an EI detector (0.60 mm pixel size at the iso-center) and a PC detector (0.25 mm pixel size). A tube voltage of 120 kV was used. Tube current-time product for EICT was 355 mAs (23.8 mGy CTDI32 cm). Dose-matched UHR-PCCT (UHRdm, 23.8 mGy) and noise-matched acquisitions (UHRnm, 10.5 mGy) were performed and reconstructed to a voxel size of 0.156 mm using a sharp kernel. Measurements of bone mineral density (BMD) and trabecular separation (Tb.Sp) and Tb.Sp percentiles reflecting the different scales of the trabecular interspacing were performed and compared to a gold-standard measurement using a peripheral CT device (XtremeCT, SCANCO Medical, Switzerland) with an isotropic voxel size of 0.082 mm and 6.6 mGy CTDI10 cm. The image noise was quantified and the relative error with respect to the gold-standard along with the agreement between CT protocols using Lin's concordance correlation coefficient (rCCC) were calculated. The Mean ±â€¯StdDev of the measured image noise levels in EICT was 109.6 ±â€¯3.9 HU. UHRdm acquisitions (same dose as EICT) showed a significantly lower noise level of 78.6 ±â€¯4.6 HU (p = 0.0122). UHRnm (44% dose of EICT) showed a noise level of 115.8 ±â€¯3.7 HU, very similar to EICT at the same spatial resolution. For BMD the overall Mean ±â€¯StdDev for EI, UHRdm and UHRnm were 114.8 ±â€¯28.6 mgHA/cm3, 121.6 ±â€¯28.8 mgHA/cm3 and 121.5 ±â€¯28.6 mgHA/cm3, respectively, compared to 123.1 ±â€¯25.5 mgHA/cm3 for XtremeCT. For Tb.Sp these values were 1.86 ±â€¯0.54 mm, 1.80 ±â€¯0.56 mm and 1.84 ±â€¯0.52 mm, respectively, compared to 1.66 ±â€¯0.48 mm for XtremeCT. The ranking of the vertebrae with regard to Tb.Sp data was maintained throughout all Tb.Sp percentiles and among the CT protocols and the gold-standard. The agreement between protocols was very good for all comparisons: UHRnm vs. EICT (BMD rCCC = 0.97; Tb.Sp rCCC = 0.998), UHRnm vs. UHRdm (BMD rCCC = 0.998; Tb.Sp rCCC = 0.993) and UHRdm vs. EICT (BMD rCCC = 0.97; Tb.Sp rCCC = 0.991). Consequently, the relative RMS-errors from linear regressions against the gold-standard for EICT, UHRdm and UHRnm were very similar for BMD (7.1%, 5.2% and 5.4%) and for Tb.Sp (3.3%, 3.3% and 2.9%), with a much lower radiation dose for UHRnm. Short-term reproducibility for BMD measurements was similar and below 0.2% for all protocols, but for Tb.Sp showed better results for UHR (about 1/3 of the level for EICT). In conclusion, CT with UHR-PC detectors demonstrated lower image noise and better reproducibility for assessments of bone microstructure at similar dose levels. For UHRnm, radiation exposure levels could be reduced by 56% without deterioration of performance levels in the assessment of bone mineral density and bone microstructure.

Dorsal Root Ganglion Morphometric Changes Under Oxaliplatin Treatment : Longitudinal Assessment by Computed Tomography.

PURPOSE: Magnetic resonance neurography (MRN) can detect dorsal root ganglia (DRG) hypertrophy in patients with oxaliplatin-induced peripheral neuropathy (OXIPN) but is difficult to apply in clinical daily practice. Aims of this study were (i) to assess whether DRG volume is reliably measurable by routine computed tomography (CT) scans, (ii) to measure longitudinal changes in DRG during and after oxaliplatin administration and (iii) to assess correlation between DRG morphometry and individual oxaliplatin dose. METHODS: For comparison of MRN and CT measurements, CT scans of 18 patients from a previous MRN study were analyzed. For longitudinal assessment of DRG size under treatment, 96 patients treated with oxaliplatin between January and December 2014 were enrolled retrospectively. DRG volumetry was performed by analyzing routine CT scans, starting with the last scan before oxaliplatin exposure (t0) and up to four consecutive timepoints after initiation of oxaliplatin therapy (t1-t4) with the following median and ranges in months: 3.1 (0.4-4.9), 6.2 (5.3-7.8), 10.4 (8.2-11.9), and 18.4 (12.8-49.8). RESULTS: DRG volume measured in CT showed a moderately strong correlation with MRN (r = 0.51, p < 0.001) and a strong correlation between two consecutive CTs (r = 0.77, p < 0.001). DRG volume increased after oxaliplatin administration with a maximum at timepoint t2. Higher cumulative oxaliplatin exposure was associated with significantly higher absolute DRG volumes (p = 0.005). Treatment discontinuation was associated with a nonsignificant trend towards lower relative DRG volume changes (p = 0.08). CONCLUSION: CT is a reliable method for continuous DRG morphometry; however, since no standardized assessment of OXIPN was performed in this retrospective study, correlations between DRG size, cumulative oxaliplatin dose and clinical symptoms in future prospective studies are needed to establish DRG size as a potential OXIPN biomarker.

Assessment of Sodium MRI at 7 Tesla as Predictor of Therapy Response and Survival in Glioblastoma Patients.

The purpose of this work was to prospectively investigate sodium (23Na) MRI at 7 Tesla (T) as predictor of therapy response and survival in patients with glioblastoma (GBM). Thus, 20 GBM patients underwent 23Na MRI at 7T before, immediately after and 6 weeks after chemoradiotherapy (CRT). The median tissue sodium concentration (TSC) inside the whole tumor excluding necrosis was determined. Initial response to CRT was assessed employing the updated response assessment in neuro-oncology working group (RANO) criteria. Clinical parameters, baseline TSC and longitudinal TSC differences were compared between patients with initial progressive disease (PD) and patients with initial stable disease (SD) using Fisher's exact tests and Mann-Whitney-U-tests. Univariate proportional hazard models for progression free survival (PFS) and overall survival (OS) were calculated using clinical parameters and TSC metrics as predictor variables. The analyses demonstrated that TSC developed heterogeneously over all patients following CRT. None of the TSC metrics differed significantly between cases of initial SD and initial PD. Furthermore, TSC metrics did not yield a significant association with PFS or OS. Conversely, the initial response according to the RANO criteria could significantly predict PFS [univariate HR (95%CI) = 0.02 (0.0001-0.21), p < 0.001] and OS [univariate HR = 0.17 (0.04-0.65), p = 0.005]. In conclusion, TSC showed treatment-related changes in GBM following CRT, but did not significantly correlate with the initial response according to the RANO criteria, PFS or OS. In contrast, the initial response according to the RANO criteria was a significant predictor of PFS and OS. Future investigations need to elucidate the reasons for treatment-related changes in TSC and their clinical value for response prediction in glioblastoma patients receiving CRT.

Improvement of PI-RADS-dependent prostate cancer classification by quantitative image assessment using radiomics or mean ADC.

Background Currently, interpretation of prostate MRI is performed qualitatively. Quantitative assessment of the mean apparent diffusion coefficient (mADC) is promising to improve diagnostic accuracy while radiomic machine learning (RML) allows to probe complex parameter spaces to identify the most promising multi-parametric models. We have previously developed quantitative RML and ADC classifiers for prediction of clinically significant prostate cancer (sPC) from prostate MRI, however these have not been combined with radiologist PI-RADS assessment. Purpose To propose and evaluate diagnostic algorithms combining quantitative ADC or RML and qualitative PI-RADS assessment for prediction of sPC. Methods and population The previously published quantitative models (RML and mADC) were utilized to construct four algorithms: 1) Down(ADC) and 2) Down(RML): clinically detected PI-RADS positive prostate lesions (defined as either PI-RADS≥3 or ≥4) were downgraded to MRI negative upon negative quantitative assessment; and 3) Up(ADC) and 4) Up(RML): MRI-negative lesions were upgraded to MRI-positive upon positive assessment of quantitative parameters. Analyses were performed at the individual lesion level and the patient level in 133 consecutive patients with suspicion for clinically significant prostate cancer (sPC, International Society of Urological Pathology (ISUP) grade group≥2), the test set subcohort of a previously published patient population. McNemar test was used to compare differences in sensitivity, specificity and accuracy. Differences between lesions of different prostate zones were assessed using ANOVA. Reduction in false positive assessments was assessed as ratios. Results Compared to clinical assessment at the PI-RADS≥4 cut-off alone, algorithms Down(ADC/RML) improved specificity from 43% to 65% (p = 0.001)/62% (p = 0.003), while sensitivity did not change significantly at 89% compared to 87% (p = 1.0)/89% (unchanged) on the patient level. Reduction of false positive lesions was 50% [26/52] in the PZ and 53% [15/28] in the TZ. Algorithms Up(ADC/RML) led, on a patient basis, to an unfavorable loss of specificity from 43% to 30% (p = 0.039)/32% (p = 0.106), with insignificant increase of sensitivity from 89% to 96%/96% (both p = 1.0). Compared to clinical assessment at the PI-RADS≥3 cut-off alone, similar results were observed for Down(ADC) with significantly increased specificity from 2% to 23% (p < 0.001) and unchanged sensitivity on the lesion level; patient level specificity increased only non-significantly. Conclusion Downgrading PI-RADS≥3 and ≥ 4 lesions based on quantitative mADC measurements or RML classifiers can increase diagnostic accuracy by enhancing specificity and preserving sensitivity for detection of sPC and reduce false positives.

Measured Multipoint Ultra-High b-Value Diffusion MRI in the Assessment of MRI-Detected Prostate Lesions.

OBJECTIVES: The aim of this study was to assess quantitative ultra-high b-value (UHB) diffusion magnetic resonance imaging (MRI)-derived parameters in comparison to standard clinical apparent diffusion coefficient (SD-ADC-2b-1000, SD-ADC-2b-1500) for the prediction of clinically significant prostate cancer, defined as Gleason Grade Group greater than or equal to 2. MATERIALS AND METHODS: Seventy-three patients who underwent 3-T prostate MRI with diffusion-weighted imaging acquired at b = 50/500/1000/1500s/mm2 and b = 100/500/1000/1500/2250/3000/4000 s/mm2 were included. Magnetic resonance lesions were segmented manually on individual sequences, then matched to targeted transrectal ultrasonography/MRI fusion biopsies. Monoexponential 2-point and multipoint fits of standard diffusion and of UHB diffusion were calculated with incremental b-values. Furthermore, a kurtosis fit with parameters Dapp and Kapp with incremental b-values was obtained. Each parameter was examined for prediction of clinically significant prostate cancer using bootstrapped receiver operating characteristics and decision curve analysis. Parameter models were compared using Vuong test. RESULTS: Fifty of 73 men (age, 66 years [interquartile range, 61-72]; prostate-specific antigen, 6.6 ng/mL [interquartile range, 5-9.7]) had 64 MRI-detected lesions. The performance of SD-ADC-2b-1000 (area under the curve, 0.82) and SD-ADC-2b-1500 (area under the curve, 0.82) was not statistically different (P = 0.99), with SD-ADC-2b-1500 selected as reference. Compared with the reference model, none of the 19 tested logistic regression parameter models including multipoint and 2-point UHB-ADC, Dapp, and Kapp with incremental b-values of up to 4000 s/mm2 outperformed SD-ADC-2b-1500 (all P's > 0.05). Decision curve analysis confirmed these results indicating no higher net benefit for UHB parameters in comparison to SD-ADC-2b-1500 in the clinically important range from 3% to 20% of cancer threshold probability. Net reduction analysis showed no reduction of MR lesions requiring biopsy. CONCLUSIONS: Despite evaluation of a large b-value range and inclusion of 2-point, multipoint, and kurtosis models, none of the parameters provided better predictive performance than standard 2-point ADC measurements using b-values 50/1000 or 50/1500. Our results suggest that most of the diagnostic benefits available in diffusion MRI are already represented in an ADC composed of one low and one 1000 to 1500 s/mm2 b-value.

Simulated clinical deployment of fully automatic deep learning for clinical prostate MRI assessment.

OBJECTIVES: To simulate clinical deployment, evaluate performance, and establish quality assurance of a deep learning algorithm (U-Net) for detection, localization, and segmentation of clinically significant prostate cancer (sPC), ISUP grade group ≥ 2, using bi-parametric MRI. METHODS: In 2017, 284 consecutive men in active surveillance, biopsy-naïve or pre-biopsied, received targeted and extended systematic MRI/transrectal US-fusion biopsy, after examination on a single MRI scanner (3 T). A prospective adjustment scheme was evaluated comparing the performance of the Prostate Imaging Reporting and Data System (PI-RADS) and U-Net using sensitivity, specificity, predictive values, and the Dice coefficient. RESULTS: In the 259 eligible men (median 64 [IQR 61-72] years), PI-RADS had a sensitivity of 98% [106/108]/84% [91/108] with a specificity of 17% [25/151]/58% [88/151], for thresholds at ≥ 3/≥ 4 respectively. U-Net using dynamic threshold adjustment had a sensitivity of 99% [107/108]/83% [90/108] (p > 0.99/> 0.99) with a specificity of 24% [36/151]/55% [83/151] (p > 0.99/> 0.99) for probability thresholds d3 and d4 emulating PI-RADS ≥ 3 and ≥ 4 decisions respectively, not statistically different from PI-RADS. Co-occurrence of a radiological PI-RADS ≥ 4 examination and U-Net ≥ d3 assessment significantly improved the positive predictive value from 59 to 63% (p = 0.03), on a per-patient basis. CONCLUSIONS: U-Net has similar performance to PI-RADS in simulated continued clinical use. Regular quality assurance should be implemented to ensure desired performance. KEY POINTS: • U-Net maintained similar diagnostic performance compared to radiological assessment of PI-RADS ≥ 4 when applied in a simulated clinical deployment. • Application of our proposed prospective dynamic calibration method successfully adjusted U-Net performance within acceptable limits of the PI-RADS reference over time, while not being limited to PI-RADS as a reference. • Simultaneous detection by U-Net and radiological assessment significantly improved the positive predictive value on a per-patient and per-lesion basis, while the negative predictive value remained unchanged.

Assessment of Melanin Content and its Influence on Susceptibility Contrast in Melanoma Metastases.

PURPOSE: To quantify the influence of melanin content on magnetic susceptibility of cerebral melanoma metastases. METHODS: Patients with non-hemorrhagic metastases were included based on the absence of susceptibility blooming artifacts. Susceptibility maps were calculated from 3D gradient echo data, using Laplacian-based phase unwrapping, sophisticated harmonic artefact reduction for phase data (V-SHARP) with varying spherical kernel sizes for background field removal and the iLSQR algorithm for the inversion of phase data. Susceptibility maps were referenced to cerebrospinal fluid. Non-hemorrhagic metastases were identified on contrast-enhanced T1-weighted images and susceptibility weighted images. Metastases masks were drawn on T1-weighted post-contrast images and used to compute mean susceptibility values of each metastasis. RESULTS: A total of 33 non-hemorrhagic melanoma brain metastases in 20 patients were quantitatively evaluated. Metastases without and with hyperintense signal on T1-weighted images, which corresponds to the melanin content, showed median susceptibility values of -0.028 ppm and -0.020 ppm, respectively. The susceptibility differences between metastases without and with T1-weighted hyperintense signal was not statistically significant (p ≥ 0.05). CONCLUSION: Non-hemorrhagic cerebral melanoma metastases showed weak diamagnetic susceptibility values and susceptibility did not significantly correlate to T1-weighted signals. Therefore, melanin does not seem to be a major contributor to susceptibility in cerebral melanoma metastases.

Classification of Cancer at Prostate MRI: Deep Learning versus Clinical PI-RADS Assessment.

Background Men suspected of having clinically significant prostate cancer (sPC) increasingly undergo prostate MRI. The potential of deep learning to provide diagnostic support for human interpretation requires further evaluation. Purpose To compare the performance of clinical assessment to a deep learning system optimized for segmentation trained with T2-weighted and diffusion MRI in the task of detection and segmentation of lesions suspicious for sPC. Materials and Methods In this retrospective study, T2-weighted and diffusion prostate MRI sequences from consecutive men examined with a single 3.0-T MRI system between 2015 and 2016 were manually segmented. Ground truth was provided by combined targeted and extended systematic MRI-transrectal US fusion biopsy, with sPC defined as International Society of Urological Pathology Gleason grade group greater than or equal to 2. By using split-sample validation, U-Net was internally validated on the training set (80% of the data) through cross validation and subsequently externally validated on the test set (20% of the data). U-Net-derived sPC probability maps were calibrated by matching sextant-based cross-validation performance to clinical performance of Prostate Imaging Reporting and Data System (PI-RADS). Performance of PI-RADS and U-Net were compared by using sensitivities, specificities, predictive values, and Dice coefficient. Results A total of 312 men (median age, 64 years; interquartile range [IQR], 58-71 years) were evaluated. The training set consisted of 250 men (median age, 64 years; IQR, 58-71 years) and the test set of 62 men (median age, 64 years; IQR, 60-69 years). In the test set, PI-RADS cutoffs greater than or equal to 3 versus cutoffs greater than or equal to 4 on a per-patient basis had sensitivity of 96% (25 of 26) versus 88% (23 of 26) at specificity of 22% (eight of 36) versus 50% (18 of 36). U-Net at probability thresholds of greater than or equal to 0.22 versus greater than or equal to 0.33 had sensitivity of 96% (25 of 26) versus 92% (24 of 26) (both P > .99) with specificity of 31% (11 of 36) versus 47% (17 of 36) (both P > .99), not statistically different from PI-RADS. Dice coefficients were 0.89 for prostate and 0.35 for MRI lesion segmentation. In the test set, coincidence of PI-RADS greater than or equal to 4 with U-Net lesions improved the positive predictive value from 48% (28 of 58) to 67% (24 of 36) for U-Net probability thresholds greater than or equal to 0.33 (P = .01), while the negative predictive value remained unchanged (83% [25 of 30] vs 83% [43 of 52]; P > .99). Conclusion U-Net trained with T2-weighted and diffusion MRI achieves similar performance to clinical Prostate Imaging Reporting and Data System assessment. © RSNA, 2019 Online supplemental material is available for this article. See also the editorial by Padhani and Turkbey in this issue.

Automated quantitative tumour response assessment of MRI in neuro-oncology with artificial neural networks: a multicentre, retrospective study.

BACKGROUND: The Response Assessment in Neuro-Oncology (RANO) criteria and requirements for a uniform protocol have been introduced to standardise assessment of MRI scans in both clinical trials and clinical practice. However, these criteria mainly rely on manual two-dimensional measurements of contrast-enhancing (CE) target lesions and thus restrict both reliability and accurate assessment of tumour burden and treatment response. We aimed to develop a framework relying on artificial neural networks (ANNs) for fully automated quantitative analysis of MRI in neuro-oncology to overcome the inherent limitations of manual assessment of tumour burden. METHODS: In this retrospective study, we compiled a single-institution dataset of MRI data from patients with brain tumours being treated at Heidelberg University Hospital (Heidelberg, Germany; Heidelberg training dataset) to develop and train an ANN for automated identification and volumetric segmentation of CE tumours and non-enhancing T2-signal abnormalities (NEs) on MRI. Independent testing and large-scale application of the ANN for tumour segmentation was done in a single-institution longitudinal testing dataset from the Heidelberg University Hospital and in a multi-institutional longitudinal testing dataset from the prospective randomised phase 2 and 3 European Organisation for Research and Treatment of Cancer (EORTC)-26101 trial (NCT01290939), acquired at 38 institutions across Europe. In both longitudinal datasets, spatial and temporal tumour volume dynamics were automatically quantified to calculate time to progression, which was compared with time to progression determined by RANO, both in terms of reliability and as a surrogate endpoint for predicting overall survival. We integrated this approach for fully automated quantitative analysis of MRI in neuro-oncology within an application-ready software infrastructure and applied it in a simulated clinical environment of patients with brain tumours from the Heidelberg University Hospital (Heidelberg simulation dataset). FINDINGS: For training of the ANN, MRI data were collected from 455 patients with brain tumours (one MRI per patient) being treated at Heidelberg hospital between July 29, 2009, and March 17, 2017 (Heidelberg training dataset). For independent testing of the ANN, an independent longitudinal dataset of 40 patients, with data from 239 MRI scans, was collected at Heidelberg University Hospital in parallel with the training dataset (Heidelberg test dataset), and 2034 MRI scans from 532 patients at 34 institutions collected between Oct 26, 2011, and Dec 3, 2015, in the EORTC-26101 study were of sufficient quality to be included in the EORTC-26101 test dataset. The ANN yielded excellent performance for accurate detection and segmentation of CE tumours and NE volumes in both longitudinal test datasets (median DICE coefficient for CE tumours 0·89 [95% CI 0·86-0·90], and for NEs 0·93 [0·92-0·94] in the Heidelberg test dataset; CE tumours 0·91 [0·90-0·92], NEs 0·93 [0·93-0·94] in the EORTC-26101 test dataset). Time to progression from quantitative ANN-based assessment of tumour response was a significantly better surrogate endpoint than central RANO assessment for predicting overall survival in the EORTC-26101 test dataset (hazard ratios ANN 2·59 [95% CI 1·86-3·60] vs central RANO 2·07 [1·46-2·92]; p<0·0001) and also yielded a 36% margin over RANO (p<0·0001) when comparing reliability values (ie, agreement in the quantitative volumetrically defined time to progression [based on radiologist ground truth vs automated assessment with ANN] of 87% [266 of 306 with sufficient data] compared with 51% [155 of 306] with local vs independent central RANO assessment). In the Heidelberg simulation dataset, which comprised 466 patients with brain tumours, with 595 MRI scans obtained between April 27, and Sept 17, 2018, automated on-demand processing of MRI scans and quantitative tumour response assessment within the simulated clinical environment required 10 min of computation time (average per scan). INTERPRETATION: Overall, we found that ANN enabled objective and automated assessment of tumour response in neuro-oncology at high throughput and could ultimately serve as a blueprint for the application of ANN in radiology to improve clinical decision making. Future research should focus on prospective validation within clinical trials and application for automated high-throughput imaging biomarker discovery and extension to other diseases. FUNDING: Medical Faculty Heidelberg Postdoc-Program, Else Kröner-Fresenius Foundation.

Early response assessment of glioma patients to definitive chemoradiotherapy using chemical exchange saturation transfer imaging at 7 T.

BACKGROUND: Patients with newly diagnosed inoperable glioma receive chemoradiotherapy (CRT). Standard Response Assessment in Neuro-Oncology (RANO) takes a minimum of 4 weeks after the end of treatment. PURPOSE/HYPOTHESIS: To investigate whether chemical exchange saturation transfer (CEST) MRI enables earlier assessment of response to CRT in glioma patients. STUDY TYPE: Longitudinal prospective study. POPULATION: Twelve brain tumor patients who underwent definitive CRT were included in this study. Three longitudinal CEST MRI measurements were performed for each patient at 7T: first before, second immediately after completion of CRT, and a third measurement as a 6-week follow-up. FIELD STRENGTH/SEQUENCE: Conventional MRI (contrast-enhanced, T2 w and diffusion-weighted imaging) at 3T and T2 w and CEST MRI at 7T was performed for all patients. ASSESSMENT: The mean relaxation-compensated relayed nuclear-Overhauser-effect CEST signal (rNOE) and the mean downfield-rNOE-suppressed amide proton transfer (dns-APT) CEST signal were investigated. Additionally, choline-to-N-acetyl-aspartate ratios (Cho/NAA) were evaluated using single-voxel 1 H-MRS in six of these patients. Performance of obtained contrasts was analyzed in assessing treatment response as classified according to the updated RANO criteria. STATISTICAL TEST: Unpaired Student's t-test. RESULTS: The rNOE signal significantly separated stable and progressive disease directly after the end of therapy (post-treatment normalized to pre-treatment mean ± SD: rNOEresponder = 1.090 ± 0.110, rNOEnon-responder = 0.808 ± 0.155, P = 0.015). In contrast, no significant difference was observed between either group when assessing the normalized dns-APT (dns-APTresponder = 0.953 ± 0.384, dns-APTnon-responder = 0.972 ± 0.477, P = 0.95). In the smaller MRS subcohort, normalized Cho/NAA decreased in therapy responders (Cho/NAAresponder = 0.632 ± 0.007, Cho/NAAnon-responder = 0.946 ± 0.124, P = 0.070). DATA CONCLUSION: rNOE mediated CEST imaging at 7T allowed for discrimination of responders and non-responders immediately after the end of CRT, additionally supported by 1 H-MRS data. This is at least 4 weeks earlier than the standard clinical evaluation according to RANO. Therefore, CEST MRI may enable early response assessment in glioma patients. LEVEL OF EVIDENCE: 1 Technical Efficacy Stage: 5 J. Magn. Reson. Imaging 2019;50:1268-1277.

Guidelines for Acquisition, Interpretation, and Reporting of Whole-Body MRI in Myeloma: Myeloma Response Assessment and Diagnosis System (MY-RADS).

Acknowledging the increasingly important role of whole-body MRI for directing patient care in myeloma, a multidisciplinary, international, and expert panel of radiologists, medical physicists, and hematologists with specific expertise in whole-body MRI in myeloma convened to discuss the technical performance standards, merits, and limitations of currently available imaging methods. Following guidance from the International Myeloma Working Group and the National Institute for Clinical Excellence in the United Kingdom, the Myeloma Response Assessment and Diagnosis System (or MY-RADS) imaging recommendations are designed to promote standardization and diminish variations in the acquisition, interpretation, and reporting of whole-body MRI in myeloma and allow response assessment. This consensus proposes a core clinical protocol for whole-body MRI and an extended protocol for advanced assessments. Published under a CC BY 4.0 license. Online supplemental material is available for this article.

Global Challenges for Cancer Imaging.

Imaging plays many essential roles in nearly all aspects of high-quality cancer care. However, challenges to the delivery of optimal cancer imaging in both developing and advanced countries are manifold. Developing countries typically face dramatic shortages of both imaging equipment and general radiologists, and efforts to improve cancer imaging in these countries are often complicated by poor infrastructure, cultural barriers, and other obstacles. In advanced countries, on the other hand, although imaging equipment and general radiologists are typically accessible, the complexity of oncologic imaging and the need for subspecialists in the field are largely unrecognized; as a result, training opportunities are lacking, and there is a shortage of radiologists with the necessary subspecialty expertise to provide optimal cancer care and participate in advanced clinical research. This article is intended to raise awareness of these challenges and catalyze further efforts to address them. Some promising strategies and ongoing efforts are reviewed, and some specific actions are proposed.

Sensitive and non-invasive assessment of hepatocellular iron using a novel room-temperature susceptometer.

BACKGROUND & AIMS: Liver iron accumulates in various chronic liver diseases where it is an independent factor for survival and carcinogenesis. We tested a novel room-temperature susceptometer (RTS) to non-invasively assess liver iron concentration (LIC). METHODS: Two hundred and sixty-four patients with or without signs of iron overload or liver disease were prospectively enrolled. Thirty-five patients underwent liver biopsy with semiquantitative iron determination (Prussian Blue staining), atomic absorption spectroscopy (AAS, n=33), or magnetic resonance imaging (MRI, n=15). RESULTS: In vitro studies demonstrated a highly linear (r2=0.998) association between RTS-signal and iron concentration, with a detection limit of 0.3mM. Using an optimized algorithm, accounting for the skin-to-liver capsule distance, valid measurements could be obtained in 84% of cases. LIC-RTS showed a significant correlation with LIC-AAS (r=0.74, p<0.001), LIC-MRI (r=0.64, p<0.001) and hepatocellular iron (r=0.58, p<0.01), but not with macrophage iron (r=0.32, p=0.30). Normal LIC-RTS was 1.4mg/g dry weight. Besides hereditary and transfusional iron overload, LIC-RTS was also significantly elevated in patients with alcoholic liver disease. The areas under the receiver operating characteristic curve (AUROC) for grade 1, 2 and 3 hepatocellular iron overload were 0.72, 0.89 and 0.97, respectively, with cut-off values of 2.0, 4.0 and 5.0mg/g dry weight. Notably, the positive and negative predictive values, sensitivity, specificity and accuracy of severe hepatic iron overload (HIO) (grade ≥2) detection, were equal to AAS and superior to all serum iron markers. Depletion of hepatic iron could be efficiently monitored upon phlebotomy. CONCLUSIONS: RTS allows for the rapid and non-invasive measurement of LIC. In comparison to MRI, it could be a cost-effective bedside method for LIC screening. Lay summary: Novel room-temperature susceptometer (RTS) allows for the rapid, sensitive, and non-invasive measurement of liver iron concentration. In comparison to MRI, it could be a cost-effective bedside method for liver iron concentration screening.

Assessment of tumor oxygenation and its impact on treatment response in bevacizumab-treated recurrent glioblastoma.

Antiantiogenic therapy with bevacizumab in recurrent glioblastoma is currently understood to both reduce microvascular density and to prune abnormal tumor microvessels. Microvascular pruning and the resulting vascular normalization are hypothesized to reduce tumor hypoxia and increase supply of systemic therapy to the tumor; however, the underlying pathophysiological changes and their timing after treatment initiation remain controversial. Here, we use a novel dynamic susceptibility contrast MRI-based method, which allows simultaneous assessment of tumor net oxygenation changes reflected by the tumor metabolic rate of oxygen and vascular normalization represented by the capillary transit time heterogeneity. We find that capillary transit time heterogeneity, and hence the oxygen extraction fraction combine with the tumoral blood flow (cerebral blood flow) in such a way that the overall tumor oxygenation appears to be worsened despite vascular normalization. Accordingly, hazards for both progression and death are found elevated in patients with a greater reduction of tumor metabolic rate of oxygen in response to bevacizumab and patients with higher intratumoral tumor metabolic rate of oxygen at baseline. This implies that tumors with a higher degree of angiogenesis prior to bevacizumab-treatment retain a higher level of angiogenesis during therapy despite a greater antiangiogenic effect of bevacizumab, hinting at evasive mechanisms limiting bevacizumab efficacy in that a reversal of their biological behavior and relative prognosis does not occur.

New bone post-processing tools in forensic imaging: a multi-reader feasibility study to evaluate detection time and diagnostic accuracy in rib fracture assessment.

PURPOSE: The aim of this multi-reader feasibility study was to evaluate new post-processing CT imaging tools in rib fracture assessment of forensic cases by analyzing detection time and diagnostic accuracy. MATERIALS AND METHODS: Thirty autopsy cases (20 with and 10 without rib fractures in autopsy) were randomly selected and included in this study. All cases received a native whole body CT scan prior to the autopsy procedure, which included dissection and careful evaluation of each rib. In addition to standard transverse sections (modality A), CT images were subjected to a reconstruction algorithm to compute axial labelling of the ribs (modality B) as well as "unfolding" visualizations of the rib cage (modality C, "eagle tool"). Three radiologists with different clinical and forensic experience who were blinded to autopsy results evaluated all cases in a random manner of modality and case. RESULTS: Rib fracture assessment of each reader was evaluated compared to autopsy and a CT consensus read as radiologic reference. A detailed evaluation of relevant test parameters revealed a better accordance to the CT consensus read as to the autopsy. Modality C was the significantly quickest rib fracture detection modality despite slightly reduced statistic test parameters compared to modalities A and B. CONCLUSION: Modern CT post-processing software is able to shorten reading time and to increase sensitivity and specificity compared to standard autopsy alone. The eagle tool as an easy to use tool is suited for an initial rib fracture screening prior to autopsy and can therefore be beneficial for forensic pathologists.

Applicability and discriminative value of a semiautomatic three-dimensional spherical volume for the assessment of the apparent diffusion coefficient in suspicious breast lesions-feasibility study.

INTRODUCTION: To evaluate the feasibility and accuracy of a semiautomatic, three-dimensional volume of interest (3D sphere) for measuring the apparent diffusion coefficient (ADC) in suspicious breast lesions compared to conventional single-slice two-dimensional regions of interest (2D ROIs). METHOD: This institutional-review-board-approved study included 56 participants with Breast Imaging Reporting and Data System 4/5 lesion. All received diffusion-weighted imaging magnetic resonance imaging prior to biopsy (b=0-1500 s/mm2). ADC values were measured in the lesions with both methods. Reproducibility and accuracies were compared. RESULTS: Area under the curve was 0.93 [95% confidence interval (CI) 0.86-0.99] for the 3D sphere and 0.91 (95% CI 0.84-0.98) for the 2D ROIs without significantly differing reproducibility (P=.45). CONCLUSION: A semiautomatic 3D sphere could reliably estimate ADC values in suspicious breast lesions without significant difference compared to conventional 2D ROIs.

An efficient computational approach to model statistical correlations in photon counting x-ray detectors.

PURPOSE: To introduce and evaluate an increment matrix approach (IMA) describing the signal statistics of energy-selective photon counting detectors including spatial-spectral correlations between energy bins of neighboring detector pixels. The importance of the occurring correlations for image-based material decomposition is studied. METHODS: An IMA describing the counter increase patterns in a photon counting detector is proposed. This IMA has the potential to decrease the number of required random numbers compared to Monte Carlo simulations by pursuing an approach based on convolutions. To validate and demonstrate the IMA, an approximate semirealistic detector model is provided, simulating a photon counting detector in a simplified manner, e.g., by neglecting count rate-dependent effects. In this way, the spatial-spectral correlations on the detector level are obtained and fed into the IMA. The importance of these correlations in reconstructed energy bin images and the corresponding detector performance in image-based material decomposition is evaluated using a statistically optimal decomposition algorithm. RESULTS: The results of IMA together with the semirealistic detector model were compared to other models and measurements using the spectral response and the energy bin sensitivity, finding a good agreement. Correlations between the different reconstructed energy bin images could be observed, and turned out to be of weak nature. These correlations were found to be not relevant in image-based material decomposition. An even simpler simulation procedure based on the energy bin sensitivity was tested instead and yielded similar results for the image-based material decomposition task, as long as the fact that one incident photon can increase multiple counters across neighboring detector pixels is taken into account. CONCLUSIONS: The IMA is computationally efficient as it required about 10(2) random numbers per ray incident on a detector pixel instead of an estimated 10(8) random numbers per ray as Monte Carlo approaches would need. The spatial-spectral correlations as described by IMA are not important for the studied image-based material decomposition task. Respecting the absolute photon counts and thus the multiple counter increases by a single x-ray photon, the same material decomposition performance could be obtained with a simpler detector description using the energy bin sensitivity.