Opportunistic Prostate Cancer Screening with Biparametric Magnetic Resonance Imaging (VISIONING).

BACKGROUND: This study investigates the use of biparametric magnetic resonance imaging (bpMRI) as primary opportunistic screening for prostate cancer (PCa) without using a prostate-specific antigen (PSA) cut-off. OBJECTIVE: The primary endpoint was to assess the efforts and effectiveness of identifying 20 participants with clinically significant prostate cancer (csPCa) using bpMRI. DESIGN, SETTING, AND PARTICIPANTS: Biopsy-naïve men aged over 45 yr were included. All participants underwent 3 Tesla bpMRI, PSA, and digital rectal examination (DRE). Targeted-only biopsy was performed in participants with Prostate Imaging Reporting and Data System (PI-RADS) ≥3. Men with negative bpMRI but suspicious DRE or elevated PSA/PSA density had template biopsies. Preintended protocol adjustments were made after an interim analysis for PI-RADS 3 lesions: no biopsy and follow-up MRI after 6 mo and biopsy only if lesions persisted or upgraded. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Biopsy results underwent a comparison using Fisher's exact test and univariable logistic regression to identify prognostic factors for positive biopsy. RESULTS AND LIMITATIONS: A total of 229 men were enrolled in this study, of whom 79 underwent biopsy. Among these men, 77 displayed suspicious PI-RADS lesions. PCa was detected in 29 participants (12.7%), of whom 21 had csPCa (9.2%). Biparametric MRI detected 21 csPCa cases, while PSA and DRE would have missed 38.1%. Protocol adjustment led to a 54.6% biopsy reduction in PI-RADS 3 lesions. Overall, in this cohort of men with a median PSA value of 1.26 ng/ml, 10.9 bpMRI scans were needed to identify one participant with csPCa. A major limitation of the study is the lack of a control cohort undergoing systematic biopsies. CONCLUSIONS: Opportunistic screening utilising bpMRI as a primary tool has higher sensitivity in detecting csPCa than classical screening methods. PATIENT SUMMARY: Screening with biparametric magnetic resonance imaging (bpMRI) and targeted biopsy identified clinically significant prostate cancer in every 11th man, regardless of the prostate-specific antigen (PSA) levels. Preselecting patients based on PSA >1 ng/ml and a positive family history of prostate cancer, as well as other potential blood tests may further improve the effectiveness of bpMRI in this setting.

TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images.

Purpose: To present a deep learning segmentation model that can automatically and robustly segment all major anatomic structures on body CT images. Materials and Methods: In this retrospective study, 1204 CT examinations (from 2012, 2016, and 2020) were used to segment 104 anatomic structures (27 organs, 59 bones, 10 muscles, and eight vessels) relevant for use cases such as organ volumetry, disease characterization, and surgical or radiation therapy planning. The CT images were randomly sampled from routine clinical studies and thus represent a real-world dataset (different ages, abnormalities, scanners, body parts, sequences, and sites). The authors trained an nnU-Net segmentation algorithm on this dataset and calculated Dice similarity coefficients to evaluate the model's performance. The trained algorithm was applied to a second dataset of 4004 whole-body CT examinations to investigate age-dependent volume and attenuation changes. Results: The proposed model showed a high Dice score (0.943) on the test set, which included a wide range of clinical data with major abnormalities. The model significantly outperformed another publicly available segmentation model on a separate dataset (Dice score, 0.932 vs 0.871; P < .001). The aging study demonstrated significant correlations between age and volume and mean attenuation for a variety of organ groups (eg, age and aortic volume [rs = 0.64; P < .001]; age and mean attenuation of the autochthonous dorsal musculature [rs = -0.74; P < .001]). Conclusion: The developed model enables robust and accurate segmentation of 104 anatomic structures. The annotated dataset (https://doi.org/10.5281/zenodo.6802613) and toolkit (https://www.github.com/wasserth/TotalSegmentator) are publicly available.Keywords: CT, Segmentation, Neural Networks Supplemental material is available for this article. © RSNA, 2023See also commentary by Sebro and Mongan in this issue.

Investigating the impact of structured reporting on the linguistic standardization of radiology reports through natural language processing over a 10-year period.

OBJECTIVES: To investigate how a transition from free text to structured reporting affects reporting language with regard to standardization and distinguishability. METHODS: A total of 747,393 radiology reports dictated between January 2011 and June 2020 were retrospectively analyzed. The body and cardiothoracic imaging divisions introduced a reporting concept using standardized language and structured reporting templates in January 2016. Reports were segmented by a natural language processing algorithm and converted into a 20-dimension document vector. For analysis, dimensionality was reduced to a 2D visualization with t-distributed stochastic neighbor embedding and matched with metadata. Linguistic standardization was assessed by comparing distinct report types' vector spreads (e.g., run-off MR angiography) between reporting standards. Changes in report type distinguishability (e.g., CT abdomen/pelvis vs. MR abdomen) were measured by comparing the distance between their centroids. RESULTS: Structured reports showed lower document vector spread (thus higher linguistic similarity) compared with free-text reports overall (21.9 [free-text] vs. 15.9 [structured]; - 27.4%; p < 0.001) and for most report types, e.g., run-off MR angiography (15.2 vs. 1.8; - 88.2%; p < 0.001) or double-rule-out CT (26.8 vs. 10.0; - 62.7%; p < 0.001). No changes were observed for reports continued to be written in free text, e.g., CT head reports (33.2 vs. 33.1; - 0.3%; p = 1). Distances between the report types' centroids increased with structured reporting (thus better linguistic distinguishability) overall (27.3 vs. 54.4; + 99.3 ± 98.4%) and for specific report types, e.g., CT abdomen/pelvis vs. MR abdomen (13.7 vs. 37.2; + 171.5%). CONCLUSION: Structured reporting and the use of factual language yield more homogenous and standardized radiology reports on a linguistic level, tailored to specific reporting scenarios and imaging studies. CLINICAL RELEVANCE: Information transmission to referring physicians, as well as automated report assessment and content extraction in big data analyses, may benefit from standardized reporting, due to consistent report organization and terminology used for pathologies and normal findings. KEY POINTS: • Natural language processing and t-distributed stochastic neighbor embedding can transform radiology reports into numeric vectors, allowing the quantification of their linguistic standardization. • Structured reporting substantially increases reports' linguistic standardization (mean: - 27.4% in vector spread) and distinguishability (mean: + 99.3 ± 98.4% increase in vector distance) compared with free-text reports. • Higher standardization and homogeneity outline potential benefits of structured reporting for information transmission and big data analyses.

Hypertensive Heart Disease-The Imaging Perspective.

Hypertensive heart disease (HHD) develops in response to the chronic exposure of the left ventricle and left atrium to elevated systemic blood pressure. Left ventricular structural changes include hypertrophy and interstitial fibrosis that in turn lead to functional changes including diastolic dysfunction and impaired left atrial and LV mechanical function. Ultimately, these changes can lead to heart failure with a preserved (HFpEF) or reduced (HFrEF) ejection fraction. This review will outline the clinical evaluation of a patient with hypertension and/or suspected HHD, with a particular emphasis on the role and recent advances of multimodality imaging in both diagnosis and differential diagnosis.

Estimation of differential renal function on routine abdominal imaging employing compressed-sensed contrast-enhanced MR: a feasibility study referenced against dynamic renal scintigraphy in patients with deteriorating renal retention parameters.

PURPOSE: To assess whether high temporal/spatial resolution GRASP MRI acquired during routine clinical imaging can identify several degrees of renal function impairment referenced against renal dynamic scintigraphy. METHODS: This retrospective study consists of method development and method verification parts. During method development, patients subject to renal imaging using gadoterate meglumine and GRASP post-contrast MRI technique (TR/TE 3.3/1.6 ms; FoV320 × 320 mm; FA12°; Voxel1.1 × 1.1x2.5 mm) were matched into four equally-sized renal function groups (no-mild-moderate-severe impairment) according to their laboratory-determined estimated glomerular filtration rates (eGFR); 60|120 patients|kidneys were included. Regions-of-interest (ROIs) were placed on cortices, medullary pyramids and collecting systems of bilateral kidneys. Cortical perfusion, tubular concentration and collecting system excretion were determined as TimeCortex=Pyramid(sec), SlopeTubuli (sec-1), and TimeCollecting System (sec), respectively, and were measured by a combination of extraction of time intensity curves and respective quantitative parameters. For method verification, patients subject to GRASP MRI and renal dynamic scintigraphy (99mTc-MAG3, 100 MBq/patient) were matched into three renal function groups (no-mild/moderate-severe impairment). Split renal function parameters post 1.5-2.5 min as well as MAG3 TER were correlated with time intensity parameters retrieved using GRASP technique; 15|30 patients|kidneys were included. RESULTS: Method development showed differing values for TimeCortex=Pyramid(71|75|93|122 s), SlopeTubuli(2.6|2.1|1.3|0.5 s-1) and TimeCollecting System(90|111|129|139 s) for the four renal function groups with partial significant tendencies (several p-values < 0.001). In method verification, 29/30 kidneys (96.7%) were assigned to the correct renal function group. CONCLUSION: High temporal and spatial resolution GRASP MR imaging allows to identify several degrees of renal function impairment using routine clinical imaging with a high degree of accuracy.

Metal implants on abdominal CT: does split-filter dual-energy CT provide additional value over iterative metal artifact reduction?

PURPOSE: To assess image quality and metal artifact reduction in split-filter dual-energy CT (sfDECT) of the abdomen with hip or spinal implants using virtual monoenergetic images (VMI) and iterative metal artifact reduction algorithm (iMAR). METHODS: 102 portal-venous abdominal sfDECTs of patients with hip (n = 71) or spinal implants (n = 31) were included in this study. Images were reconstructed as 120kVp-equivalent images (Mixed) and VMI (40-190 keV), with and without iMAR. Quantitative artifact and image noise was measured using 12 different ROIs. Subjective image quality was rated by two readers using a five-point Likert-scale in six categories, including overall image quality and vascular contrast. RESULTS: Lowest quantitative artifact in both hip and spinal implants was measured in VMI190keV-iMAR. However, it was not significantly lower than in MixediMAR (for all ROIs, p = 1.00), which were rated best for overall image quality (hip: 1.00 [IQR: 1.00-2.00], spine: 3.00 [IQR:2.00-3.00]). VMI50keV-iMAR was rated best for vascular contrast (hip: 1.00 [IQR: 1.00-2.00], spine: 2.00 [IQR: 1.00-2.00]), which was significantly better than Mixed (both, p < 0.001). VMI50keV-iMAR provided superior overall image quality compared to Mixed for hip (1.00 vs 2.00, p < 0.001) and similar diagnostic image quality for spinal implants (2.00 vs 2.00, p = 0.51). CONCLUSION: For abdominal sfDECT with hip or spinal implants MixediMAR images should be used. High keV VMI do not further improve image quality. IMAR allows the use of low keV images (VMI50keV) to improve vascular contrast, compared to Mixed images.

Computed tomography (CT) and magnetic resonance imaging (MRI) of diffuse liver disease: a multiparametric predictive modelling algorithm can aid categorization of liver parenchyma.

BACKGROUND: Liver steatosis is common and tracking disease evolution to steatohepatitis and cirrhosis is essential for risk stratification and resultant patient management. Consequently, diagnostic tools allowing categorization of liver parenchyma based on routine imaging are desirable. The study objective was to compare established mono-factorial, dynamic single parameter and iterative multiparametric routine computed tomography (CT) and magnetic resonance imaging (MRI) analyses to distinguish between liver steatosis, steatohepatitis, cirrhosis and normal liver parenchyma. METHODS: A total of 285 multi-phase contrast enhanced CT and 122 MRI studies with histopathological correlation of underlying parenchymal condition were retrospectively included. Parenchymal conditions were characterized based on CT Hounsfield units (HU) or MRI signal intensity (SI) measurements and calculated HU or SI ratios between non-contrast and contrast enhanced imaging time points. First, the diagnostic accuracy of mono-factorial analyses using established, static non-contrast HU and in- to opposed phase SI change cut-offs to distinguish between parenchymal conditions was established. Second, single dynamic discriminator analyses, with optimized non-contrast and enhancement HU and SI ratio cut-off values derived from the data, employing receiver operating characteristic (ROC) curve areas under the curve (AUCs) and the Youden index for maximum accuracy, were used for disease diagnosis. Third, multifactorial analyses, employing multiple non-contrast and contrast enhanced HU and SI ratio cut-offs in a nested, predictive-modelling algorithm were performed to distinguish between normal parenchyma, liver steatosis, steatohepatitis and cirrhosis. CT and MRI analyses were performed separately. RESULTS: No single CT or MRI parameter showed significant difference between all four parenchymal conditions (each P>0.05). Mono-factorial static-CT-discriminator analyses identified liver steatosis with 75% accuracy. Mono-factorial MRI analyses identified steatosis with 89% accuracy. Single-dynamic CT parameter analyses identified normal parenchyma with 72% accuracy and cirrhosis with 75% accuracy. Single-dynamic MRI parameter analyses identified fatty parenchyma with 90% accuracy. Multifactorial CT analyzes identified normal parenchyma with 84%, liver steatosis with 95%, steatohepatitis with 95% and cirrhosis with 80% accuracy. Multifactorial predictive modelling of MRI parameters identified normal parenchyma with 79%, liver steatosis with 89%, steatohepatitis with 92% and cirrhosis with 89% accuracy. CONCLUSIONS: Multiparametric analyses of quantitative measurements derived from routine CT and MRI, utilizing a predictive modelling algorithm, can help to distinguish between normal liver parenchyma, liver steatosis, steatohepatitis and cirrhosis.

A Phase 1/2 Single-arm Clinical Trial of Recombinant Bacillus Calmette-Guérin (BCG) VPM1002BC Immunotherapy in Non-muscle-invasive Bladder Cancer Recurrence After Conventional BCG Therapy: SAKK 06/14.

BACKGROUND: VPM1002BC is a genetically modified Mycobacterium bovis bacillus Calmette-Guérin (BCG) strain with potentially improved immunogenicity and attenuation. OBJECTIVE: To report on the efficacy, safety, tolerability and quality of life of intravesical VPM1002BC for the treatment of non-muscle-invasive bladder cancer (NMIBC) recurrence after conventional BCG therapy. DESIGN, SETTING, AND PARTICIPANTS: We designed a phase 1/2 single-arm trial (NCT02371447). Patients with recurrent NMIBC after BCG induction ± BCG maintenance therapy and intermediate to high risk for cancer progression were eligible. INTERVENTION: Patients were scheduled for standard treatment of six weekly instillations with VPM1002BC followed by maintenance for 1 yr. Treatment was stopped in cases of recurrence. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: The primary endpoint was defined as the recurrence-free rate (RFR) in the bladder 60 wk after trial registration. The sample size was calculated based on the assumption that ≥30% of the patients would be without recurrence at 60 wk after registration. RESULTS AND LIMITATIONS: After exclusion of two ineligible patients, 40 patients remained in the full analysis set. All treated tumours were of high grade and 27 patients (67.5%) presented with carcinoma in situ. The recurrence-free rate in the bladder at 60 wk after trial registration was 49.3% (95% confidence interval [CI] 32.1-64.4%) and remained at 47.4% (95% CI 30.4-62.6%] at 2 yr and 43.7% (95% CI 26.9-59.4%) at 3 yr after trial registration. At the same time, progression to muscle-invasive disease had occurred in three patients and metastatic disease in four patients. Treatment-related grade 1, 2, and 3 adverse events (AEs) were observed in 14.3%, 54.8%, and 4.8% of the patients, respectively. No grade ≥4 AEs occurred. Two of the 42 patients did not tolerate five or more instillations during induction. Limitations include the single-arm trial design and the low number of patients for subgroup analysis. CONCLUSIONS: At 1 yr after treatment start, almost half of the patients remained recurrence-free after therapy with VPM100BC. The primary endpoint of the study was met and the therapy is safe and well tolerated. PATIENT SUMMARY: We conducted a trial of VPM100BC, a genetically modified bacillus Calmette-Guérin (BCG) strain for treatment of bladder cancer not invading the bladder muscle. At 1 year after the start of treatment, almost half of the patients with a recurrence after previous conventional BCG were free from non-muscle-invasive bladder cancer (NMIBC). The results are encouraging and VPM1002BC merits further evaluation in randomised studies for patients with NMIBC.

Dual-energy CT of acute bowel ischemia.

Acute bowel ischemia is a condition with high mortality and requires rapid intervention to avoid catastrophic outcomes. Swift and accurate imaging diagnosis is essential because clinical findings are commonly nonspecific. Conventional contrast enhanced CT of the abdomen has been the imaging modality of choice to evaluate suspected acute bowel ischemia. However, subtlety of image findings and lack of non-contrast or arterial phase images can make correct diagnosis challenging. Dual-energy CT provides valuable information toward assessing bowel ischemia. Dual-energy CT exploits the differential X-ray attenuation at two different photon energy levels to characterize the composition of tissues and reveal the presence or absence of faint intravenous iodinated contrast to improve reader confidence in detecting subtle bowel wall enhancement. With the same underlying technique, virtual non-contrast images can help to show non-enhancing hyperdense hemorrhage of the bowel wall in intravenous contrast-enhanced scans without the need to acquire actual non-contrast scans. Dual-energy CT derived low photon energy (keV) virtual monoenergetic images emphasize iodine contrast and provide CT angiography-like images from portal venous phase scans to better evaluate abdominal arterial patency. In Summary, dual-energy CT aids diagnosing acute bowel ischemia in multiple ways, including improving visualization of the bowel wall and mesenteric vasculature, revealing intramural hemorrhage in contrast enhanced scans, or possibly reducing intravenous contrast dose.

Supervised learning based on tumor imaging and biopsy transcriptomics predicts response of hepatocellular carcinoma to transarterial chemoembolization.

Although transarterial chemoembolization (TACE) is the most widely used treatment for intermediate-stage, unresectable hepatocellular carcinoma (HCC), it is only effective in a subset of patients. In this study, we combine clinical, radiological, and genomics data in supervised machine-learning models toward the development of a clinically applicable predictive classifier of response to TACE in HCC patients. Our study consists of a discovery cohort of 33 tumors through which we identify predictive biomarkers, which are confirmed in a validation cohort. We find that radiological assessment of tumor area and several transcriptomic signatures, primarily the expression of FAM111B and HPRT1, are most predictive of response to TACE. Logistic regression decision support models consisting of tumor area and RNA-seq gene expression estimates for FAM111B and HPRT1 yield a predictive accuracy of ∼90%. Reverse transcription droplet digital PCR (RT-ddPCR) confirms these genes in combination with tumor area as a predictive classifier for response to TACE.

Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning.

Pancreatic cystic lesions (PCL) are a frequent and underreported incidental finding on CT scans and can transform into neoplasms with devastating consequences. We developed and evaluated an algorithm based on a two-step nnU-Net architecture for automated detection of PCL on CTs. A total of 543 cysts on 221 abdominal CTs were manually segmented in 3D by a radiology resident in consensus with a board-certified radiologist specialized in abdominal radiology. This information was used to train a two-step nnU-Net for detection with the performance assessed depending on lesions' volume and location in comparison to three human readers of varying experience. Mean sensitivity was 78.8 ± 0.1%. The sensitivity was highest for large lesions with 87.8% for cysts ≥220 mm3 and for lesions in the distal pancreas with up to 96.2%. The number of false-positive detections for cysts ≥220 mm3 was 0.1 per case. The algorithm's performance was comparable to human readers. To conclude, automated detection of PCL on CTs is feasible. The proposed model could serve radiologists as a second reading tool. All imaging data and code used in this study are freely available online.

Investigating difficult to detect pancreatic lesions: Characterization of benign pancreatic islet cell tumors using multiparametric pancreatic 3-T MRI.

INTRODUCTION: Pancreatic islet-cell tumors (PICT) often present with atypical signal-characteristics and are often missed on preoperative imaging. The aim of this study is to provide a multiparametric PICT characterization and investigate factors impeding PICT detection. MATERIAL AND METHODS: This is a detailed MRI analysis of a prospective, monocenter study, including 49 consecutive patients (37 female, 12 male; median age 50) with symptoms due to endogenous hyperinsulinemic hypoglycemia (EHH) and mostly negative prior-imaging. All patients received a 3-T MRI and a 68Ga-DOTA-exendin-4-PET/CT. Pooled accuracy, sensitivity, specificity and inter-reader agreement were calculated. Reference-standard was histopathology and 68Ga-DOTA-Exendin-4-PET/CT in one patient who refused surgery. For PICT analyses, 34 patients with 49 PICTs (48 histologically proven; one 68Ga-DOTA-exendin-4-PET/CT positive) were assessed. Dynamic contrast-enhanced (DCE) Magnetic Resonance Images (MRI) with Golden-Angle-Radial-Sparse-Parallel (GRASP) reconstruction, enabling imaging at high spatial and temporal resolution, was used to assess enhancement-patterns of PICTs. Tumor-to-background (T2B) ratio for each sequence and the employed quantitative threshold for conspicuity of PICTs were analyzed in regard to prediction of true-positive PICTs. RESULTS: Evaluation of 49 patients revealed a pooled lesion-based accuracy, sensitivity and specificity of 70.3%, 72.9% and 62.5%, respectively. Mean PICT size was 12.9±5.3mm for detected, 9.0±2.9mm for undetected PICTs (p-value 0.0112). In-phase T1w detected the most PICT (67.3%). Depending on the sequence, PICTs were isointense and poorly visible in 29-68%. Only 2/41(4.9%) PICTs showed typical signal-characteristics across T1w, T2w, DWI and ceT1w combined. 66.6% of PICTs enhanced simultaneously to the parenchyma, 17.8% early and 15.6% late. Predictor screening analysis showed number of sequences detecting a PICT, lesion size and in-phase T1w T2B ratio had the highest contribution for detecting a true-positive PICT. CONCLUSION: The majority of PICTs enhance simultaneously to surrounding parenchyma, present with atypical signal-characteristics and thus are poorly visible. In non-enhancing PICTs, radiologists should search for small lesions most likely conspicuous on unenhanced T1w or DWI.

Evaluation of liver fibrosis and cirrhosis on the basis of quantitative T1 mapping: Are acute inflammation, age and liver volume confounding factors?

PURPOSE: To evaluate potential confounding factors in the quantitative assessment of liver fibrosis and cirrhosis using T1 relaxation times. METHODS: The study population is based on a radiology-information-system database search for abdominal MRI performed from July 2018 to April 2019 at our institution. After applying exclusion criteria 200 (59 ±â€¯16 yrs) remaining patients were retrospectively included. 93 patients were defined as liver-healthy, 40 patients without known fibrosis or cirrhosis, and 67 subjects had a clinically or biopsy-proven liver fibrosis or cirrhosis. T1 mapping was performed using a slice based look-locker approach. A ROI based analysis of the left and the right liver was performed. Fat fraction, R2*, liver volume, laboratory parameters, sex, and age were evaluated as potential confounding factors. RESULTS: T1 values were significantly lower in healthy subjects without known fibrotic changes (1.5 T MRI: 575 ±â€¯56 ms; 3 T MRI: 857 ±â€¯128 ms) compared to patients with acute liver disease (1.5 T MRI: 657 ±â€¯73 ms, p < 0.0001; 3 T MRI: 952 ±â€¯37 ms, p = 0.028) or known fibrosis or cirrhosis (1.5 T MRI: 644 ±â€¯83 ms, p < 0.0001; 3 T MRI: 995 ±â€¯150 ms, p = 0.018). T1 values correlated moderately with the Child-Pugh stage at 1.5 T (p = 0.01, ρ = 0.35). CONCLUSION: T1 mapping is a capable predictor for detection of liver fibrosis and cirrhosis. Especially age is not a confounding factor and, hence, age-independent thresholds can be defined. Acute liver diseases are confounding factors and should be ruled out before employing T1-relaxometry based thresholds to screen for patients with liver fibrosis or cirrhosis.

Revisiting DCE-MRI: Classification of Prostate Tissue Using Descriptive Signal Enhancement Features Derived From DCE-MRI Acquisition With High Spatiotemporal Resolution.

METHODS: A retrospective study (from January 2016 to July 2019) including 75 subjects (mean, 65 years; 46-80 years) with 2.5-second temporal resolution DCE-MRI and PIRADS 4 or 5 lesions was performed. Fifty-four subjects had biopsy-proven prostate cancer (Gleason 6, 15; Gleason 7, 20; Gleason 8, 13; Gleason 9, 6), whereas 21 subjects had negative MRI/ultrasound fusion-guided biopsies. Voxel-wise analysis of contrast signal enhancement was performed for all time points using custom-developed software, including automatic arterial input function detection. Seven descriptive parameter maps were calculated: normalized maximum signal intensity, time to start, time to maximum, time-to-maximum slope, and maximum slope with normalization on maximum signal and the arterial input function (SMN1, SMN2). The parameters were compared with ADC using multiparametric machine-learning models to determine classification accuracy. A Wilcoxon test was used for the hypothesis test and the Spearman coefficient for correlation. RESULTS: There were significant differences (P < 0.05) for all 7 DCE-derived parameters between the normal peripheral zone versus PIRADS 4 or 5 lesions and the biopsy-positive versus biopsy-negative lesions. Multiparametric analysis showed better performance when combining ADC + DCE as input (accuracy/sensitivity/specificity, 97%/93%/100%) relative to ADC alone (accuracy/sensitivity/specificity, 94%/95%/95%) and to DCE alone (accuracy/sensitivity/specificity, 78%/79%/77%) in differentiating the normal peripheral zone from PIRADS lesions, biopsy-positive versus biopsy-negative lesions (accuracy/sensitivity/specificity, 68%/33%/81%), and Gleason 6 versus ≥7 prostate cancer (accuracy/sensitivity/specificity, 69%/60%/72%). CONCLUSIONS: Descriptive perfusion characteristics derived from high-resolution DCE-MRI using model-free computations show significant differences between normal and cancerous tissue but do not reach the accuracy achieved with solely ADC-based classification. Combining ADC with DCE-based input features improved classification accuracy for PIRADS lesions, discrimination of biopsy-positive versus biopsy-negative lesions, and differentiation between Gleason 6 versus Gleason ≥7 lesions.

A Novel Deep Learning Based Computer-Aided Diagnosis System Improves the Accuracy and Efficiency of Radiologists in Reading Biparametric Magnetic Resonance Images of the Prostate: Results of a Multireader, Multicase Study.

OBJECTIVE: The aim of this study was to evaluate the effect of a deep learning based computer-aided diagnosis (DL-CAD) system on radiologists' interpretation accuracy and efficiency in reading biparametric prostate magnetic resonance imaging scans. MATERIALS AND METHODS: We selected 100 consecutive prostate magnetic resonance imaging cases from a publicly available data set (PROSTATEx Challenge) with and without histopathologically confirmed prostate cancer. Seven board-certified radiologists were tasked to read each case twice in 2 reading blocks (with and without the assistance of a DL-CAD), with a separation between the 2 reading sessions of at least 2 weeks. Reading tasks were to localize and classify lesions according to Prostate Imaging Reporting and Data System (PI-RADS) v2.0 and to assign a radiologist's level of suspicion score (scale from 1-5 in 0.5 increments; 1, benign; 5, malignant). Ground truth was established by consensus readings of 3 experienced radiologists. The detection performance (receiver operating characteristic curves), variability (Fleiss κ), and average reading time without DL-CAD assistance were evaluated. RESULTS: The average accuracy of radiologists in terms of area under the curve in detecting clinically significant cases (PI-RADS ≥4) was 0.84 (95% confidence interval [CI], 0.79-0.89), whereas the same using DL-CAD was 0.88 (95% CI, 0.83-0.94) with an improvement of 4.4% (95% CI, 1.1%-7.7%; P = 0.010). Interreader concordance (in terms of Fleiss κ) increased from 0.22 to 0.36 (P = 0.003). Accuracy of radiologists in detecting cases with PI-RADS ≥3 was improved by 2.9% (P = 0.10). The median reading time in the unaided/aided scenario was reduced by 21% from 103 to 81 seconds (P < 0.001). CONCLUSIONS: Using a DL-CAD system increased the diagnostic accuracy in detecting highly suspicious prostate lesions and reduced both the interreader variability and the reading time.

Enhancing disease awareness for tuberous sclerosis complex in patients with radiologic diagnosis of renal angiomyolipoma: an observational study.

BACKGROUND: Tuberous Sclerosis Complex (TSC) is a genetic disorder, with renal manifestations like angiomyolipoma (AML) occurring in 70-80% of patients. AML usually cause more complications in TCS patients than in non-TSC patients. However, AML patients are not routinely investigated for TSC. Our aim was to retrospectively assess the correlation between radiologically diagnosed AML and TSC. METHODS: All patients were stratified into AML related vs. unrelated to TSC. Correlations were calculated to determine the association between age, AML, and TSC. RESULTS: Complete data were available for 521 patients with renal AML, in 7 of which the concurrent diagnosis of TSC was found. Younger age significantly positively correlated with the prevalence of TSC in AML patients (p <  0.01). 37 (7%) of the 521 patients were within the age-range of 18-40 years, in which TSC occurred in 6 cases, 4 (66.7%) of which presented with multiple, bilateral renal AML (p <  0.05), and 2 (33.3%) of which with a single, unilateral AML (p <  0.05). In patients with AML but without TSC, unilateral AML was found in 83.9% and bilateral AML in 16.1% (p <  0.05). Simple binary logistic regression analysis revealed bilateral AML (OR 33.0; 95% CI 3.2-344.0; p = 0.003) (but not unilateral AML (OR 0.09; 95% CI 0.01-0.88; p = 0.04)) to be a risk factor for TSC. CONCLUSIONS: The presence of bilateral AML in patients within the age-range of 18-40 years should raise suspicion for TSC as the underlying cause. Therefore, our advice is to refer patients with multiple bilateral renal AML for further investigations regarding TSC.

Quantifying Radiology Resident Fatigue: Analysis of Preliminary Reports.

Background Workloads in radiology departments have constantly increased over the past decades. The resulting radiologist fatigue is considered a rising problem that affects diagnostic accuracy. Purpose To investigate whether data mining of quantitative parameters from the report proofreading process can reveal daytime and shift-dependent trends in report similarity as a surrogate marker for resident fatigue. Materials and Methods Data from 117 402 radiology reports written by residents between September 2017 and March 2020 were extracted from a report comparison tool and retrospectively analyzed. Through calculation of the Jaccard similarity coefficient between residents' preliminary and staff-reviewed final reports, the amount of edits performed by staff radiologists during proofreading was quantified on a scale of 0 to 1 (1: perfect similarity, no edits). Following aggregation per weekday and shift, data were statistically analyzed by using simple linear regression or one-way analysis of variance (significance level, P < .05) to determine relationships between report similarity and time of day and/or weekday reports were dictated. Results Decreasing report similarity with increasing work hours was observed for day shifts (r = -0.93 [95% CI: -0.73, -0.98]; P < .001) and weekend shifts (r = -0.72 [95% CI: -0.31, -0.91]; P = .004). For day shifts, negative linear correlation was strongest on Fridays (r = -0.95 [95% CI: -0.80, -0.99]; P < .001), with a 16% lower mean report similarity at the end of shifts (0.85 ± 0.24 at 8 am vs 0.69 ± 0.32 at 5 pm). Furthermore, mean similarity of reports dictated on Fridays (0.79 ± 0.35) was lower than that on all other weekdays (range, 0.84 ± 0.30 to 0.86 ± 0.27; P < .001). For late shifts, report similarity showed a negative correlation with the course of workweeks, showing a continuous decrease from Monday to Friday (r = -0.98 [95% CI: -0.70, -0.99]; P = .007). Temporary increases in report similarity were observed after lunch breaks (day and weekend shifts) and with the arrival of a rested resident during overlapping on-call shifts. Conclusion Decreases in report similarity over the course of workdays and workweeks suggest aggravating effects of fatigue on residents' report writing performances. Periodic breaks within shifts potentially foster recovery. © RSNA, 2021.

Revealing the most common reporting errors through data mining of the report proofreading process.

OBJECTIVES: To investigate the most common errors in residents' preliminary reports, if structured reporting impacts error types and frequencies, and to identify possible implications for resident education and patient safety. MATERIAL AND METHODS: Changes in report content were tracked by a report comparison tool on a word level and extracted for 78,625 radiology reports dictated from September 2017 to December 2018 in our department. Following data aggregation according to word stems and stratification by subspecialty (e.g., neuroradiology) and imaging modality, frequencies of additions/deletions were analyzed for findings and impression report section separately and compared between subgroups. RESULTS: Overall modifications per report averaged 4.1 words, with demonstrably higher amounts of changes for cross-sectional imaging (CT: 6.4; MRI: 6.7) than non-cross-sectional imaging (radiographs: 0.2; ultrasound: 2.8). The four most frequently changed words (right, left, one, and none) remained almost similar among all subgroups (range: 0.072-0.117 per report; once every 9-14 reports). Albeit representing only 0.02% of analyzed words, they accounted for up to 9.7% of all observed changes. Subspecialties solely using structured reporting had substantially lower change ratios in the findings report section (mean: 0.2 per report) compared with prose-style reporting subspecialties (mean: 2.0). Relative frequencies of the most changed words remained unchanged. CONCLUSION: Residents' most common reporting errors in all subspecialties and modalities are laterality discriminator confusions (left/right) and unnoticed descriptor misregistration by speech recognition (one/none). Structured reporting reduces overall error rates, but does not affect occurrence of the most common errors. Increased error awareness and measures improving report correctness and ensuring patient safety are required. KEY POINTS: • The two most common reporting errors in residents' preliminary reports are laterality discriminator confusions (left/right) and unnoticed descriptor misregistration by speech recognition (one/none). • Structured reporting reduces the overall the error frequency in the findings report section by a factor of 10 (structured reporting: mean 0.2 per report; prose-style reporting: 2.0) but does not affect the occurrence of the two major errors. • Staff radiologist review behavior noticeably differs between radiology subspecialties.

Response prediction of hepatocellular carcinoma undergoing transcatheter arterial chemoembolization: unlocking the potential of CT texture analysis through nested decision tree models.

OBJECTIVES: To investigate if nested multiparametric decision tree models based on tumor size and CT texture parameters from pre-therapeutic imaging can accurately predict hepatocellular carcinoma (HCC) lesion response to transcatheter arterial chemoembolization (TACE). MATERIALS AND METHODS: This retrospective study (January 2011-September 2017) included consecutive pre- and post-therapeutic dynamic CT scans of 37 patients with 92 biopsy-proven HCC lesions treated with drug-eluting bead TACE. Following manual segmentation of lesions according to modified Response Evaluation Criteria in Solid Tumors criteria on baseline arterial phase CT images, tumor size and quantitative texture parameters were extracted. HCCs were grouped into lesions undergoing primary TACE (VT-lesions) or repeated TACE (RT-lesions). Distinct multiparametric decision tree models to predict complete response (CR) and progressive disease (PD) for the two groups were generated. AUC and model accuracy were assessed. RESULTS: Thirty-eight of 72 VT-lesions (52.8%) and 8 of 20 RT-lesions (40%) achieved CR. Sixteen VT-lesions (22.2%) and 8 RT-lesions (40%) showed PD on follow-up imaging despite TACE treatment. Mean of positive pixels (MPP) was significantly higher in VT-lesions compared to RT-lesions (180.5 vs 92.8, p = 0.001). The highest AUC in ROC curve analysis and accuracy was observed for the prediction of CR in VT-lesions (AUC 0.96, positive predictive value 96.9%, accuracy 88.9%). Prediction of PD in VT-lesions (AUC 0.88, accuracy 80.6%), CR in RT-lesions (AUC 0.83, accuracy 75.0%), and PD in RT-lesions (AUC 0.86, accuracy 80.0%) was slightly inferior. CONCLUSIONS: Nested multiparametric decision tree models based on tumor heterogeneity and size can predict HCC lesion response to TACE treatment with high accuracy. They may be used as an additional criterion in the multidisciplinary treatment decision-making process. KEY POINTS: • HCC lesion response to TACE treatment can be predicted with high accuracy based on baseline tumor heterogeneity and size. • Complete response of HCC lesions undergoing primary TACE was correctly predicted with 88.9% accuracy and a positive predictive value of 96.9%. • Progressive disease was correctly predicted with 80.6% accuracy for lesions undergoing primary TACE and 80.0% accuracy for lesions undergoing repeated TACE.

Autonomous Detection and Classification of PI-RADS Lesions in an MRI Screening Population Incorporating Multicenter-Labeled Deep Learning and Biparametric Imaging: Proof of Concept.

BACKGROUND: Opportunistic prostate cancer (PCa) screening is a controversial topic. Magnetic resonance imaging (MRI) has proven to detect prostate cancer with a high sensitivity and specificity, leading to the idea to perform an image-guided prostate cancer (PCa) screening; Methods: We evaluated a prospectively enrolled cohort of 49 healthy men participating in a dedicated image-guided PCa screening trial employing a biparametric MRI (bpMRI) protocol consisting of T2-weighted (T2w) and diffusion weighted imaging (DWI) sequences. Datasets were analyzed both by human readers and by a fully automated artificial intelligence (AI) software using deep learning (DL). Agreement between the algorithm and the reports-serving as the ground truth-was compared on a per-case and per-lesion level using metrics of diagnostic accuracy and k statistics; Results: The DL method yielded an 87% sensitivity (33/38) and 50% specificity (5/10) with a k of 0.42. 12/28 (43%) Prostate Imaging Reporting and Data System (PI-RADS) 3, 16/22 (73%) PI-RADS 4, and 5/5 (100%) PI-RADS 5 lesions were detected compared to the ground truth. Targeted biopsy revealed PCa in six participants, all correctly diagnosed by both the human readers and AI. CONCLUSIONS: The results of our study show that in our AI-assisted, image-guided prostate cancer screening the software solution was able to identify highly suspicious lesions and has the potential to effectively guide the targeted-biopsy workflow.