PURPOSE:Evaluate global interest in middle meningeal artery (MMA) embolization, for subdural hematoma, by analyzing search behavior and academic output using Google Trends and PubMed data. MATERIALS AND METHODS:Retrospective analysis conducted using Google Trends data from January 2004 to March 2025 and PubMed publication records from 1965 to 2025. The Google dataset included normalized monthly search volume indices (SVI: 0-100), geographic distribution, and related queries. Academic interest was assessed by annual publication counts. Statistical methods included Mann-Kendall trend tests, Augmented Dickey-Fuller (ADF) unit root tests, Portmanteau white noise tests, and Spearman correlations to evaluate significance and associations. Forecasts were generated using polynomial regression models. RESULTS:Search interest remained negligible from 2004 to 2011, increasing modestly through 2019 before accelerating significantly. PubMed publications followed a similar pattern, with sharp growth beginning in 2015. Both datasets exhibited significant upward trends (Mann-Kendall: Tau = 0.80 for PubMed; 0.75 for Google, P < .000001). Trends were confirmed as deterministic (ADF P < .001) and non-random (Portmanteau P < .001). A strong correlation was observed between academic and public interest (Spearman r = .86, R2 = .74, P < .001). Polynomial forecasting (R2 = .93) projects continued publication growth, estimating 267 ± 24 publications by 2027. Search interest is expected to stabilize at high levels (SVI 70-90). CONCLUSION:Interest in MMA embolization is rapidly accelerating across both public and academic domains. These trends suggest a paradigm shift that may soon influence treatment guidelines, care delivery models, and the global adoption of this technique.
BACKGROUND:Physicians in Canada are disproportionately concentrated in urban centers, with rural regions facing significant gaps in specialist care. The geo-spatial characteristics of Interventional Radiology (IR) in Canada have not been systematically studied. This study examines the distribution of Canadian Association of Interventional Radiology (CAIR) members and the sociodemographic characteristics of the populations they serve. METHODS:A list of active members of the Canadian Association of IR (CAIR) was obtained in May 2025. Data from the Canadian Census was used to gather census division (CD)-level information on population size and demographics including data on ethnicity, income, education, employment, and income. CDs were grouped and compared between those with access to IR care and those without, compared between top and bottom quartiles for each population characteristic, and mapped. RESULTS:Of 293 CDs in Canada, only 48 (19.6%) had at least one CAIR IR, leaving 31.6% of the national population without local access. CAIR IRs were strongly clustered in high-population CDs, with Toronto, Vancouver, and Montreal alone accounting for 56.5% of IRs. CDs with CAIR IRs had significantly higher mean populations (527 322 vs 47 675; P < .001), median incomes ($82 760 vs $75 821; P = .002), and educational attainment (31% vs 16%; P < .001). Conversely, CDs with higher proportions of the population being Indigenous were less likely to have CAIR IRs (5% vs 14%; P < .001). CONCLUSION:CAIR member IRs are disproportionately concentrated in large urban centers, leaving substantial geographic disparities. Addressing this maldistribution will require coordinated strategies, including outreach models, telehealth integration, and national workforce planning.
Despite the centrality of radiology and radiologists in much clinical investigation and care of patients, the precise roles radiologists fulfil are often poorly understood by patients, and, sometimes, sadly, by our colleagues in healthcare. Loose use of language often designates those who refer patients for radiological investigation as "clinicians," implying that this descriptor does not apply to radiologists. The term "clinician" applies to physicians with direct responsibility for patient care; this includes typical radiologist activity. For a variety of reasons (personal preference, workload demands, modern methods of communication and working), direct contact between radiologists and patients, and between radiologists and referring colleagues is in decline. Yet this trend is occurring against a backdrop of increased reliance upon radiology and other investigative tests, as opposed to history-taking and physical examination, for diagnosis and management decision-making. Given the opportunity this trend creates for radiology and radiologists to demonstrate and enhance our clinical centrality, and the concurrent increasing expectation on the part of patients for direct interaction with those making decisions with respect to their healthcare, radiologists need to emphasise our clinical role by embracing every opportunity for direct contact with our patients and our referring colleagues, rather than allowing ourselves to become overwhelmed by and hidden behind ever-increasing workload. It is only by being true, active clinicians that radiologists can deliver most value for our patients, and ensure that our speciality prospers and grows.
Patient-reported outcome measures (PROMs) are standardized, validated instruments that measure how patients feel and function, collected directly from the patient. Traditionally, key metrics in radiology include technical aspects such as image quality, radiation dose, and diagnostic accuracy. However, medical imaging and image-guided therapies shape patient experience in informational, emotional, physical, and logistical domains that are rarely measured. Failing to capture this information is an important gap in radiology research and practice today that needs to be addressed. This review synthesizes the science of PROMs through a radiology lens: what PROMs are; why PROMs are relevant to diagnostic imaging and interventional practice; how to select and interpret PROMs responsibly, with explicit attention to bias, conflicts of interest, and minimal important differences; and how to implement PROMs pragmatically using contemporary digital workflows. This article highlights radiology-specific frameworks for patient-centred outcomes of diagnostic tests, summarizes evidence on how electronic PROM (ePROM) programs can improve patient experience and clinical outcomes, and proposes a practical roadmap for department-level implementation. Throughout, this review aligns recommendations with current methodological and regulatory guidance, draws on Canadian implementation experience, and translates lessons from applied PROM programs in complex clinical services to radiology settings. Implemented thoughtfully, PROMs give radiologists a rigorous, low-burden way to document benefits radiology already provides, strengthen outcome and health-economic analyses, and co-design services around what patients value. Integrating PROMs alongside established technical and diagnostic metrics can extend radiology's value proposition, and make radiology's patient-centred impact visible, measurable, and improvable.
Diffuse liver diseases, including steatosis, cirrhosis, and fibrosis, significantly alter hepatic architecture, and can obscure, mimic, or alter the imaging features of focal liver lesions and reduce the diagnostic accuracy of specific imaging modalities. This review outlines the key imaging pitfalls created by diffuse liver disease and offers practical, evidence-based strategies to overcome them. We describe several key imaging pitfalls in interpreting liver lesions in diffuse liver disease, including the atypical appearance of focal lesions on ultrasound and CT due to altered background parenchyma; the phenomenon of "vanishing washout" of hepatocellular carcinoma on CT in steatotic livers; and the significant risk of underestimating metastatic disease burden or missing occult lesions. For each pitfall, we discuss the strengths and weaknesses of various imaging modalities and suggest problem-solving strategies, such as guidance on selecting the most appropriate next-line imaging modality for lesion characterization. We highlight the roles of contrast-enhanced ultrasound and MRI in resolving diagnostic uncertainty when initial ultrasound or CT imaging is atypical or equivocal. A comprehensive understanding of how diffuse liver disease affects the imaging of both benign and malignant lesions is important for accurate diagnosis. Recognizing these pitfalls allows a tailored, multimodality imaging approach that optimizes the detection and characterization of focal liver lesions, ultimately guiding clinical management and improving patient care.
Acute non-traumatic pelvic pain represents a frequent and complex diagnostic dilemma in pre-menopausal females presenting to the emergency department due to overlapping symptomatology across pregnancy and non-pregnancy related causes. Radiologists play a pivotal role in expediting accurate diagnosis and guiding appropriate management in these potentially life-threatening scenarios. This review provides an approach to workup and imaging selection in these patients, emphasizing the necessity of serum β-hCG testing and the central role of transabdominal and transvaginal pelvic ultrasound. An overview of female pelvic anatomy is provided. Using a multimodality imaging approach, early pregnancy related complications such as ectopic pregnancy, retained products of conception, and gestational trophoblastic disease and non-pregnancy related causes of acute pelvic pain such as ovarian torsion, ruptured ovarian cysts, pelvic inflammatory disease, endometriosis, uterine vascular malformation, ovarian vein thrombosis, ovarian hyperstimulation syndrome, and intrauterine device complications are reviewed. Finally, we propose an algorithmic approach to imaging selection and interpretation tailored to the clinical scenario, laboratory findings (notably β-hCG status), and patient demographics. This structured framework aims to support radiologists in efficiently narrowing the differential diagnosis and optimizing patient outcomes in acute, non-traumatic pelvic emergencies.
This review provides a comprehensive overview of recent transformative advancements in diagnostic imaging that position Japan at the forefront of radiological innovation. We highlight pivotal innovations that enhance diagnostic capabilities and redefine clinical workflows. The article begins with upright multidetector computed tomography (MDCT), a groundbreaking technology offering novel insights into posture-dependent anatomical and physiological changes. We then explore significant progress in breast and gynecologic imaging, including advancements in artificial intelligence computer-aided (AI-CAD) synthesized mammograms, automated breast ultrasound (ABUS), and abbreviated MRI protocols. These innovations address unique regional challenges in early cancer detection. Significant innovations in abdominal radiology, spanning advanced CT (including photon-counting detector CT), accelerated MRI, and AI applications, are also discussed. The review further delves into glymphatic system research, where advanced MRI techniques, particularly DTI-ALPS, are unraveling new insights into brain waste clearance and neurological disorders. Finally, we examine the future of Japanese radiology through the lens of AI, with a focus on Large Language Models (LLMs). We discuss their growing role in diagnostic support, report generation, and information extraction, as well as important societal and ethical considerations. These collective advancements underscore Japan's dynamic contributions to radiological innovation, poised to significantly impact global healthcare practices by improving disease detection, optimizing workflows, and extending healthy life expectancy in an aging society.
PURPOSE:Osteoporosis is an under-screened musculoskeletal disorder that results in diminished quality of life and significant burden to the healthcare system. We aimed to evaluate the ability of Rho, an artificial intelligence (AI) tool, to prospectively identify patients at-risk for low bone mineral density (BMD) from standard x-rays, its adoption rate by radiologists, and acceptance by primary care providers (PCPs). METHODS:Patients ≥50 years were recruited when undergoing an x-ray of a Rho-eligible body part for any clinical indication. Questionnaires were completed at baseline and 6-month follow-up, and PCPs of "Rho-Positive" patients (those likely to have low BMD) were asked for feedback. Positive predictive value (PPV) was calculated in patients who returned within 6 months for a DXA. RESULTS:Of 1145 patients consented, 987 had x-rays screened by Rho, and 655 were flagged as Rho-Positive. Radiologists included this finding in 524 (80%) of reports. Of all Rho-Positive patients, 125 had a DXA within 6 months; Rho had a 74% PPV for DXA T-Score <-1. From 51 PCP responses, 78% found Rho beneficial. Of 389 patients with follow-up questionnaire data, a greater proportion of patients with Rho results reported vs not reported by a radiologist had discussed bone health with their PCP since study start (37% vs 21%, P < .001), or were newly diagnosed with osteoporosis (12% vs 5%; P = .02). CONCLUSION:By identifying patients at-risk of low BMD, with acceptability of reporting by radiologists and generally positive feedback from PCPs, Rho has the potential to improve low screening rates for osteoporosis by leveraging existing x-ray data.
PURPOSE:This study examines factors contributing to burnout among radiology residents through a Canadian lens and assesses strategies employed at our institution to mitigate its impact. METHODS:This was a single-institution cross-sectional study. Four anonymous online surveys were administered through Qualtrics to PGY 2-5 radiology residents from 2021 to 2025. These surveys identified residents with burnout and distress and assessed contributing factors, suggestions for reducing burnout, and residents' responses to implemented interventions. Interventions were employed at 2 hospitals within our institution. RESULTS:The surveys had response rates of 30% (2021), 57.7% (2023), 60% (2024), and 62% (2025). 50% of pre-intervention respondents were identified as burned out. The rate reduced to 18.8% post-intervention, with results not being statistically significant (P = .167). Top factors driving burnout included time (eg, increased work hours, time constraints), extra duties (clinical and administrative), and perceived lack of radiology knowledge when dealing with complex cases. Interventions included additional daily 1-hour teaching sessions, wellness lunch rounds, debriefing sessions, transitioning from paper-based protocolling to a hybrid-electronic paper-based system, call schedule modifications, improved ergonomics, and social functions, including incorporating indoor and outdoor activities. Interventions targeting work hours were subjectively the most well-received in combating burnout. CONCLUSION:This study underscores the prevalence of burnout among radiology residents. Our institution has implemented a multi-faceted approach to address burnout within our radiology residency program.
OBJECTIVE:To assess the operational and environmental benefits of using an abbreviated protocol for hepatocellular carcinoma (HCC) surveillance. METHODS:This IRB-approved retrospective single-center quality improvement study evaluated time, energy use, and appointment access. Inclusion criteria were HCC surveillance MRIs with either a full or abbreviated imaging protocol. Exclusion criteria were other abdominopelvic MR protocols or incomplete studies. DICOM time data were extracted via Quantivly and validated with 10 prospective time studies. Exam times from PACS images were cross-referenced with DICOM data to identify and resolve extraction outliers. Power logs from 10 exams per protocol were used to quantify energy and greenhouse gas emissions. Schedule logs assessed appointment volume changes. Mean times (±SD) and energy (±SD) were reported, and Welch's t-test determined statistical significance (P < .05). RESULTS:Exam times for 487 MRIs (318 abbreviated, 169 full protocol) were analyzed, with 67 excluded. The mean duration of exam time for the abbreviated protocol was 12.0 minutes (SD: 4.3), compared with 29.7 minutes (SD: 8.8) for the full protocol (mean difference, 17.7 minutes; P < .0001). The mean energy for the abbreviated protocol was 4.7 kWh (SD: 0.6), compared with 11.7 kWh (SD: 1.3) for the full protocol (mean difference, 7.0 kWh; P < .0001). Across 318 abbreviated exams, estimated savings totaled 2226 kWh and 1494.6 kg CO2eq. Despite time savings, MRI appointment volume and scanner access remained unchanged. CONCLUSION:Abbreviated HCC surveillance MRIs cut scan time, energy use, and carbon emissions by 60%, but scheduling complexities precluded increased MRI appointments.
OBJECTIVES:Domain shift has been shown to have a major detrimental effect on AI model performance however prior studies on domain shift for MRI prostate cancer segmentation have been limited to small, or heterogenous cohorts. Our objective was to assess whether prostate cancer segmentation models trained on local MRI data continue to outperform those trained on external data with cohorts exceeding 1000. METHODS:We simulated a multi-institutional consortium using the public PICAI dataset (PICAI-TRAIN: 1241 exams, PICAI-TEST: 259) and a local dataset (LOCAL-TRAIN: 1400 exams, LOCAL-TEST: 308). IRB approval was obtained and consent waived. We compared nnUNet-v2 models trained on the combined data (CENTRAL-TRAIN) and separately on PICAI-TRAIN and LOCAL-TRAIN. Accuracy was evaluated using the open source PICAI Score on LOCAL-TEST. Significance was tested using bootstrapping. RESULTS:Just 22% (309/1400) of LOCAL-TRAIN exams would be sufficient to match the performance of a model trained on PICAI-TRAIN. The CENTRAL-TRAIN performance was similar to LOCAL-TRAIN performance, with PICAI Scores [95% CI] of 65 [58-71] and 66 [60-72], respectively. Both of these models exceeded the model trained on PICAI-TRAIN alone which had a score of 58 [51-64] (P < .002). Reducing training set size did not alter these relative trends. CONCLUSION:Domain shift limits MRI prostate cancer segmentation performance even when training with over 1000 exams from 3 external institutions. Use of local data is paramount at these scales.
PURPOSE:This study aimed to assess the experiences and challenges faced by Canadian radiologists participating in Multidisciplinary Team rounds (MDT), with a focus on demographics, meeting characteristics, preparation processes, and perceptions of workload and compensation. METHODS:The Canadian Association of Radiologists constituted a working group which developed a 35-question survey that was distributed to 1958 radiologists and radiology trainees across Canada. The survey garnered 129 complete responses, for a response rate of 6.6%. RESULTS:Respondents predominantly practiced in academic settings (65.9%) and had subspecialty training (96.1%). The majority reported that MDT rounds lasted 30 to 60 minutes and discussed 6 to 10 cases. Most radiologists (62.8%) were the sole presenters of imaging. Preparation time was often limited, with only 6.2% having dedicated time for preparation. 59.8% of respondents reported receiving additions to caseloads within the 24-hour period prior to the meetings. While 93.8% valued the opportunity for interaction with colleagues, 93.8% felt inadequately compensated for their efforts by their practices, while 92.3% felt inadequately compensated by their province. CONCLUSIONS:While most radiologists indicated adequate time for discussion and meaningful clinical engagement during rounds, many highlighted repeated challenges such as last-minute case additions, lack of protected preparation time, and technological barriers. Systemic barriers also play a role and include lack of provincial remuneration and workforce issues which in turn impact individual workload pressures.
INTRODUCTION:This study aimed to determine the positive predictive value (PPV) of magnetic resonance imaging-transrectal ultrasound (MRI-TRUS) machine fusion prostate biopsies, and to identify factors associated with a positive biopsy. METHODS:With ethics approval, we retrospectively evaluated all MRI-TRUS machine fusion prostate biopsies at our institution from September 2022 to April 2025. True positive clinically significant prostate cancers (csPCa) were defined as Gleason ≥7. PPVs were calculated overall and for PI-RADS 3, 4 and 5 categories. A generalized linear mixed model (GLMM) was created evaluating the following factors as fixed effects: PI-RADS category; prostate-specific antigen (PSA) density (<0.10, 0.10-0.15, ≥0.15 ng/mL2); lesion size (<7, 7-15, ≥15 mm); lesion location (peripheral vs transition zone); ultrasound correlate (present/absent); prostate size (<60 vs ≥60 mL); interval from MRI to biopsy (<6 months or not); and biopsy operator (2 radiologists). Referring urologist (n = 19) and reporting radiologist (n = 8) were included as random effects. RESULTS:372 patients (mean age, 67 ± 7 years) with 529 lesions underwent biopsy. The overall PPV was 314/529 (59.4%). For PI-RADS 3 to 5, PPVs were 32/72 (44.4%), 123/243 (50.6%), and 159/214 (74.3%), respectively. In GLMM analysis, PI-RADS 5 versus 3 (OR 3.6, 95% CI, 1.7-7.4), PSA density ≥0.15 ng/mL2 (OR 2.2, 95% CI, 1.2-3.8), and presence of an ultrasound correlate (OR 2.7, 95% CI, 1.7-4.2) were associated with true positive biopsies. Small lesion size <7 mm was associated with a false positive biopsy (OR 0.4, 95% CI, 0.2-0.8). CONCLUSION:The yield of fusion prostate biopsies at our institution is high. PI-RADS 5, PSA density ≥0.15 ng/mL2, and an ultrasound correlate at biopsy were associated with csPCa, whereas sub-7 mm lesions were negatively associated with csPCa.
PURPOSE:To correlate parathyroid scintigraphy results with patient biochemistry, surgery, and pathology to inform appropriateness criteria and assess biochemical metrics in the imaging workup of hyperparathyroidism. METHODS:This retrospective study included 421 patients who underwent parathyroid scintigraphy. Patients were grouped based on primary versus secondary hyperparathyroidism, and clinical profiles were reviewed for scan result, blood work, surgical results, and pathology. Performance metrics of scintigraphy were analyzed. Demographics, bloodwork, and location were compared between positive and negative scans. Predictors of positive scans were identified by multivariate logistic regression analysis. The performance of biochemistry to predict scan results was evaluated by ROC analyses. RESULTS:Positive tests-occurring in 52% of patients-were associated with higher parathyroid hormone (PTH) and corrected calcium. However, PTH was only predictive of a positive test in patients with secondary hyperparathyroidism. On multivariate analysis, male sex, corrected calcium, and younger age were predictors of a positive scan. Corrected calcium was the most predictive with an OR of 1.28 for every 0.1 mmol/L increase. Based on ROC analysis, corrected calcium had an AUC of 0.628 and a cutoff of 2.65 mmol/L maximized sensitivity (88%) and specificity (35%) for a positive test. CONCLUSION:In this large retrospective cohort, several biochemical metrics, including corrected calcium levels, were predictive of a positive scintigraphy study. Furthermore, biochemistry, including PTH levels, significantly differed between primary and secondary hyperparathyroidism suggesting that tailored biochemical metrics are required. This work sets a foundation for the development of a robust biochemical scoring system to optimize patient selection for parathyroid imaging.
PURPOSE:Accurate vessel measurement is essential in endovascular thrombectomy (EVT) for acute ischemic stroke. Discrepancies between computed tomography angiography (CTA) and digital subtraction angiography (DSA) may impact procedural planning and device selection. This study compares vessel diameter measurements from CTA and DSA in patients with middle cerebral artery (MCA) M1 occlusions. METHODS:In this single-center retrospective study, 90 consecutive patients who underwent EVT for MCA M1 occlusions between February 2020 and March 2024 were included. Vessel diameters were independently measured by 3 neuroradiologists using CTA and DSA (pre- and post-intervention). Statistical analysis included Wilcoxon signed-rank tests, intraclass correlation coefficient (ICC), and Bland-Altman analysis. RESULTS:CTA consistently overestimated vessel diameter compared to DSA. The mean M1 diameter was 2.29 ± 0.27 mm on CTA and 2.16 ± 0.30 mm on pre-EVT DSA (P < .001), with a median difference of 0.4 mm (IQR: 0.2-0.6 mm). In 70% of cases, CTA values exceeded DSA. Bland-Altman analysis confirmed a mean difference of +0.13 mm (limits of agreement: -0.25 to +0.51 mm). No significant change was observed between pre- and post-EVT DSA measurements (P = .103). Clot-side M1 segments were significantly smaller than contralateral measurements on CTA (P = .003). Inter-rater agreement was good (ICC = .785). CONCLUSIONS:CTA overestimates MCA M1 diameter relative to DSA. While the discrepancy is modest, it may influence device selection in borderline cases. Awareness of this variability is important, and further research is warranted to explore its clinical implications.
Artificial intelligence (AI) and sustainability have been the subject of much research in the field of radiology throughout 2025. The future of the radiologist and our planet has been called into question. We have had to shift focus and evolve to embrace the progress AI can bring while limiting our environmental impact and maximising efficiency in the face of an ever-increasing workload. This year's Canadian Association of Radiologists Journal "Year in Review" revisits the 10 most powerful articles published by our journal in 2025, exploring what's next for AI, sustainability and system efficiency.
PURPOSE:In pelvic MRI, Turbo Spin Echo (TSE) pulse sequences are used for T2-weighted imaging. However, its lengthy acquisition time increases the potential for artifacts. Deep learning (DL) reconstruction achieves reduced scan times without the degradation in image quality associated with other accelerated techniques. Unfortunately, a comprehensive assessment of DL-reconstruction in pelvic MRI has not been performed. The objective of this prospective study was to compare the performance of DL-TSE and conventional TSE pulse sequences in a broad spectrum of pelvic MRI indications. METHODS:Fifty-five subjects (33 females and 22 males) were scanned at 3 T using DL-TSE and conventional TSE sequences in axial and/or oblique acquisition planes. Two radiologists independently assessed image quality in 6 categories: edge definition, vessel margin sharpness, T2 Contrast Dynamic Range, artifacts, overall image quality, and lesion features. The contrast ratio was calculated for quantitative assessment. A two-tailed sign test was used for assessment. RESULTS:The 2 readers found DL-TSE to deliver equal or superior image quality than conventional TSE in most cases. There were only 3 instances out of 24 where conventional TSE was scored as providing better image quality. Readers agreed on DL-TSE superiority/inferiority/equivalence in 67% of categories in the axial plane and 75% in the oblique plane. DL-TSE also demonstrated a better contrast ratio in 75% of cases. DL-TSE reduced scan time by approximately 50%. CONCLUSION:DL-accelerated TSE sequences generally provide equal or better image quality in pelvic MRI than standard TSE with significantly reduced acquisition times.
PURPOSE:The precentral cortex normally demonstrates lower signal intensity compared to remainder of the neocortex on 2D fluid attenuated inversion recovery (FLAIR) images. Loss of this normal hypointensity bilaterally can be seen in patients with adult hypoxic-ischemic brain injury (HIBI). We have named this the 'FLAIR motor' sign (FMS). The performance of this sign for detection of HIBI is evaluated in this case-control study. METHODS:MRI studies of 74 consecutive patients with clinical evidence of HIBI following cardiac arrest formed the 'case' group. Controls comprised of normal MRI studies of an equal number of age and gender matched patients. Two fellowship-trained neuro-radiologists reviewed the MRI studies in a blinded randomized fashion and recorded the presence or absence of 'FLAIR motor' sign. RESULTS:Average time from cardiac arrest to MRI was 7.12 days (range: 1-25 days). The average sensitivity and specificity of 'FLAIR motor' sign for HIBI was 86.49% and 100% respectively. The sign demonstrated excellent inter-reader agreement (kappa >0.8). CONCLUSION:The loss of the normal hypointensity in bilateral pre-central cortex on 2D-FLAIR images is a specific and reliable MRI sign of HIBI in the subacute phase following cardiac arrest in adults.
Rationale and Objectives: Artificial intelligence (AI) tools, particularly generative models, are increasingly used to depict clinical roles in healthcare. This study evaluates whether generative AI systems accurately differentiate between radiologists and medical radiation technologists (MRTs), 2 roles often confused by patients and providers. Materials and Methods: We assessed 1380 images and videos generated by 8 text-to-image/video AI models. Five raters evaluated task-role accuracy, attire, equipment, lighting, isolation, and demographics. Statistical tests compared differences across models and roles. Results: MRTs were depicted accurately in 82.0% of outputs, while only 56.2% of radiologist images/videos were role-appropriate. Among inaccurate radiologist depictions, 79.1% misrepresented MRTs tasks. Radiologists were more often male (73.8%) and White (79.7%), while MRTs were more diverse. Stethoscope misuse, lack of disability/religious markers, and overuse of business attire for radiologists further reflected bias. Conclusion: Generative AI frequently misrepresents radiologist roles and demographics, reinforcing stereotypes and public confusion. Greater oversight and inclusion standards are needed to ensure equitable AI-generated healthcare content.
Introduction: This study aims to evaluate the outcomes of breast MRI-guided vacuum assisted biopsies (MVAB) performed on lesions with high T2 signal. Materials and Methods: We retrospectively collected of all MVAB performed at our institution between January 2016 and December 2021 for high T2 lesions. T2 hyperintensity was defined as equal or higher signal than normal lymph node. The correlation between various demographic and imaging characteristics and the binary pathological outcome (benign vs malignant) was evaluated. Results: In total, 174 biopsies from 165 women met the inclusion criteria and were included in the cohort. Malignancy was detected in 35 lesions (20%), most commonly ductal carcinoma in situ (DCIS, 57%), followed by infiltrating ductal carcinoma (IDC, 40%). The most common benign diagnosis was fibrocystic changes (FCC, 38%). In 19 lesions MVAB detected high-risk pathology, 3 of which were upgraded to malignancy. Older age (Mean 61 vs 54 years, P = .04), washout kinetics (29% vs 13%, P = .01), and indication for extent of disease evaluation (53% vs 32%, P = .06) were the strongest predictors of malignancy. Lesion size and morphology were not significantly associated with outcome. Conclusions: Given the considerable cancer rate, T2 hyperintensity should be used with caution as a benign indicator and not as a sole criterion for ruling out malignancy. Additional factors such as patient age, kinetic features, and MRI indication should be considered to improve diagnostic accuracy.
