Background Automation bias (the propensity for humans to favor suggestions from automated decision-making systems) is a known source of error in human-machine interactions, but its implications regarding artificial intelligence (AI)-aided mammography reading are unknown. Purpose To determine how automation bias can affect inexperienced, moderately experienced, and very experienced radiologists when reading mammograms with the aid of an artificial intelligence (AI) system. Materials and Methods In this prospective experiment, 27 radiologists read 50 mammograms and provided their Breast Imaging Reporting and Data System (BI-RADS) assessment assisted by a purported AI system. Mammograms were obtained between January 2017 and December 2019 and were presented in two randomized sets. The first was a training set of 10 mammograms, with the correct BI-RADS category suggested by the AI system. The second was a set of 40 mammograms in which an incorrect BI-RADS category was suggested for 12 mammograms. Reader performance, degree of bias in BI-RADS scoring, perceived accuracy of the AI system, and reader confidence in their own BI-RADS ratings were assessed using analysis of variance (ANOVA) and repeated-measures ANOVA followed by post hoc tests and Kruskal-Wallis tests followed by the Dunn post hoc test. Results The percentage of correctly rated mammograms by inexperienced (mean, 79.7% ± 11.7 [SD] vs 19.8% ± 14.0; P < .001; r = 0.93), moderately experienced (mean, 81.3% ± 10.1 vs 24.8% ± 11.6; P < .001; r = 0.96), and very experienced (mean, 82.3% ± 4.2 vs 45.5% ± 9.1; P = .003; r = 0.97) radiologists was significantly impacted by the correctness of the AI prediction of BI-RADS category. Inexperienced radiologists were significantly more likely to follow the suggestions of the purported AI when it incorrectly suggested a higher BI-RADS category than the actual ground truth compared with both moderately (mean degree of bias, 4.0 ± 1.8 vs 2.4 ± 1.5; P = .044; r = 0.46) and very (mean degree of bias, 4.0 ± 1.8 vs 1.2 ± 0.8; P = .009; r = 0.65) experienced readers. Conclusion The results show that inexperienced, moderately experienced, and very experienced radiologists reading mammograms are prone to automation bias when being supported by an AI-based system. This and other effects of human and machine interaction must be considered to ensure safe deployment and accurate diagnostic performance when combining human readers and AI. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Baltzer in this issue.

Partial segmental thrombosis of the corpus cavernosum (PSTCC) is a rare disease predominantly occurring in young men. Cardinal symptoms are pain and perineal swelling. Although several risk factors are described in the literature, the exact etiology of penile thrombosis remains unclear in most cases. MRI or ultrasound (US) is usually used for diagnosing this condition. We report a case of penile thrombosis after left-sided varicocele ligature in a young patient. The diagnosis was established using contrast-enhanced ultrasound (CEUS) and was confirmed by contrast-enhanced magnetic resonance imaging (ceMRI). Successful conservative treatment consisted of systemic anticoagulation using low molecular weight heparin and acetylsalicylic acid. PSTCC is a rare condition in young men and appears with massive pain and perineal swelling. In case of suspected PSTCC utilization of CEUS may be of diagnostic benefit.

With increasing spatial resolution, diffusion-weighted imaging (DWI) may be suitable for morphologic lesion characterization in breast MRI - an area that has traditionally been occupied by dynamic contrast-enhanced imaging (DCE). This investigation compared DWI with b values of 800 and 1600 s/mm² to DCE for lesion morphology assessment in high-resolution breast MRI at 3 Tesla. Multiparametric breast MRI was performed in 91 patients with 93 histopathologically proven lesions (31 benign, 62 malignant). Two radiologists independently evaluated three datasets per patient (DWI_b800; DWI_b1600; DCE) and assessed lesion visibility and BIRADS morphology criteria. Diagnostic accuracy was compared among readers and datasets using Cochran's Q test and pairwise post-hoc McNemar tests. Bland-Altman analyses were conducted for lesion size comparisons. Discrimination of carcinomas was superior compared to benign findings in both DWI_b800 and DWI_b1600 (p < 0.001) with no b value-dependent difference. Similarly, assessability of mass lesions was better than of non-mass lesions, irrespective of b value (p < 0.001). Intra-reader reliability for the analysis of morphologic BIRADS criteria among DCE and DWI datasets was at least moderate (Fleiss κ≥0.557), while at least substantial inter-reader agreement was ascertained over all assessed categories (κ≥0.776). In pairwise Bland-Altman analyses, the measurement bias between DCE and DWI_b800 was 0.7 mm, whereas the difference between DCE and DWI_b1600 was 2.8 mm. DWI_b1600 allowed for higher specificity than DCE (p = 0.007/0.062). DWI can be employed for reliable morphologic lesion characterization in high-resolution breast MRI. High b values increase diagnostic specificity, while lesion size assessment is more precise with standard 800 s/mm² images.

Objectives: This study investigated the feasibility and image quality of ultra-low-dose unenhanced abdominal CT using photon-counting detector technology and tin prefiltration. Materials and Methods: Employing a first-generation photon-counting CT scanner, eight cadaveric specimens were examined both with tin prefiltration (Sn 100 kVp) and polychromatic (120 kVp) scan protocols matched for radiation dose at three different levels: standard-dose (3 mGy), low-dose (1 mGy) and ultra-low-dose (0.5 mGy). Image quality was evaluated quantitatively by means of contrast-to-noise-ratios (CNR) with regions of interest placed in the renal cortex and subcutaneous fat. Additionally, three independent radiologists performed subjective evaluation of image quality. The intraclass correlation coefficient was calculated as a measure of interrater reliability. Results: Irrespective of scan mode, CNR in the renal cortex decreased with lower radiation dose. Despite similar mean energy of the applied x-ray spectrum, CNR was superior for Sn 100 kVp over 120 kVp at standard-dose (17.75 ± 3.51 vs. 14.13 ± 4.02), low-dose (13.99 ± 2.6 vs. 10.68 ± 2.17) and ultra-low-dose levels (8.88 ± 2.01 vs. 11.06 ± 1.74) (all p ≤ 0.05). Subjective image quality was highest for both standard-dose protocols (score 5; interquartile range 5–5). While no difference was ascertained between Sn 100 kVp and 120 kVp examinations at standard and low-dose levels, the subjective image quality of tin-filtered scans was superior to 120 kVp with ultra-low radiation dose (p < 0.05). An intraclass correlation coefficient of 0.844 (95% confidence interval 0.763–0.906; p < 0.001) indicated good interrater reliability. Conclusions: Photon-counting detector CT permits excellent image quality in unenhanced abdominal CT with very low radiation dose. Employment of tin prefiltration at 100 kVp instead of polychromatic imaging at 120 kVp increases the image quality even further in the ultra-low-dose range of 0.5 mGy.

Objectives: Digital breast tomosynthesis (DBT) can provide additional information over mammography, albeit at the cost of prolonged reading time. This study retrospectively investigated the impact of reading enhanced synthetic 6 mm slabs instead of standard 1 mm slices on interpretation time and readers performance in a diagnostic assessment centre. Methods: Three radiologists (R1-3; 6/4/2 years of breast imaging experience) reviewed 111 diagnostic DBT examinations. Two datasets were interpreted independently for each patient, with one set containing artificial-intelligence-enhanced synthetic 6 mm slabs with 3 mm overlap, while the other set comprised standard 1 mm slices. Blinded to histology and follow-up, readers noted individual BIRADS categories and diagnostic confidence while reading time was recorded. Among the 111 examinations, 70 findings were histopathologically correlated including 56 malignancies. Results: No significant difference was found between BIRADS categories assigned based on 6 mm vs 1 mm datasets (p ≥ 0.317). Diagnostic accuracy was comparable for 6 mm and 1 mm readings (R1: 87.0% vs 87.0%; R2: 86.1% vs 87.0%; R3: 80.0% vs 84.4%; p ≥ 0.125) with high interrater agreement (intraclass correlation coefficient 0.848 vs 0.865). One reader reported higher confidence with 1 mm slices (R1: p = 0.033). Reading time was substantially shorter when interpreting 6 mm slabs compared to 1 mm slices (R1: 33.5 vs 46.2; R2: 49.1 vs 64.8; R3: 39.5 vs 67.2 sec; all p < 0.001). Conclusions: Artificial-intelligence-enhanced synthetic 6 mm slabs allow for substantial interpretation time reduction in diagnostic DBT without a decrease in reader accuracy. Advances in knowledge: A simplified slab-only protocol instead of 1 mm slices may offset the higher reading time without a loss of diagnosis-relevant image information in first and second readings. Further evaluations are required regarding workflow implications, particularly in screening settings.