The Radiological Society of North America (RSNA) established its Gold Medal in 1919 and the American College of Radiology (ACR) followed in 1931, yet, these highest honors in radiology have overwhelmingly gone to men. Of the 211 ACR Gold Medalists, only 18 have been women; of the 221 RSNA Gold Medalists, only 23 have been women. Among the small fraction of women recipients, just nine women have earned both. These dual awardees, Marie Sklodowska-Curie, Edith Quimby, Alice Ettinger, Rosalyn Yalow, Kay Vydareny, Sarah Donaldson, Valerie Jackson, Anne Roberts, and Beverly Coleman, are true academic “unicorns.” Their careers span the physics of radioactivity and to sonography of an unborn child. We profiled them through virtual interviews with the five living awardees, published interviews, research contributions, and institutional records. Three themes surfaced, unprompted, across every conversation: leadership carried as an intentional responsibility, mentorship as a full cycle of being sponsored and then sponsoring others, and an enduring commitment to clinical excellence as the grounds of their academic and administrative authority. These nine women offer practical lessons and lasting inspiration for emerging leaders.
In 2022, Lung-RADS introduced a “Lung-RADS 0” category for CT findings suggestive of acute infectious or inflammatory processes. This study aimed to evaluate thoracic radiologist agreement with Lung-RADS 0 assignments and assess outcomes on follow up imaging.This retrospective review was conducted on lung cancer screening CT scans from November 2022 to July 2025. LungView (LungView, Fulton, MD, USA) software identified scans assigned Lung-RADS 0. Cases without follow-up were excluded. Thoracic radiologists reviewed the CT scans for agreement with Lung-RADS 0 and reclassified those with disagreements. Follow-up scans were assessed for resolution or progression of findings.Out of 4020 CT scans, 135 were categorized as Lung-RADS 0 (3.3%). After excluding 6 cases with no follow-up, 129 scans and their clinical reports were analyzed and reviewed independently by board certified radiologists within the thoracic division. The reviewers felt that 26.4% of the studies should have been originally reported as Lung RADS 2, 3 or 4, with the remainder (73.6%) deemed appropriately reported as Lung RADS 0. Resolution or improvement of findings occurred in 79/129 cases (61.2%), while 33.3% showed no change or worsening. Two malignancies were diagnosed at follow-up, neither of which was delayed due to the Lung-RADS 0 classification. This study highlights the need for consistent and correct application of Lung-RADS 0 category defined by ACR to avoid unnecessary shorter-term follow-ups. There was no delayed diagnosis of malignancy for those patients with nodules assigned as Lung RADS 0.
OBJECTIVES:To evaluate the computed tomography (CT) findings of liver injury after chest compressions in patients who achieved return of spontaneous circulation following cardiopulmonary resuscitation. METHODS:This retrospective single-center study included consecutive patients who underwent abdominal CT after successful resuscitation (1/2006-9/2024). CT images were retrospectively reviewed to identify liver injury attributable to chest compressions. Imaging findings, injury distribution, hemorrhagic features were analyzed. RESULTS:Among 716 resuscitated patients who underwent CT, liver injury after chest compressions was identified in 11 patients (mean age, 60.5 years ± 15.8 [standard deviation]; 4 men, 7 women). All liver injuries were detected on contrast-enhanced CT (5.3%, 11/206), whereas only one liver injury was detectable on noncontrast CT (0.1%, 1/716). Lacerations predominantly involved the left lobe (64%) and caudate lobe (55%), with segment 4 being the most frequent dominant site. Right lobe involvement was uncommon. Bleeding along the falciform ligament was observed in 36% of cases. Hemoperitoneum occurred in two patients, both of whom died from hemorrhagic shock related to liver injury. CONCLUSION:Liver injury after chest compressions is an uncommon but clinically relevant complication in resuscitated patients. Contrast-enhanced CT is crucial for detection. Awareness of characteristic imaging features-including preferential involvement of the left and caudate lobes, and bleeding along the falciform ligament-can improve diagnostic recognition on post-resuscitation CT.
Purpose The increasing cost of acute stroke evaluation has been well documented. Given the expanded use of intravenous thrombolysis (IVT) and endovascular thrombectomy (EVT) in published guidelines, we hypothesized that increasing imaging-associated expenditures would correspond to higher rates of reperfusion therapy. Methods A six-year retrospective analysis of all Code Stroke activations at our institution was conducted. For each Code Stroke, data on initial neuroimaging modality ordered and decision to treat with IVT or EVT was documented.Analysis evaluated trends in estimated total imaging cost, number of Code Stroke activations per year, cumulative radiation exposure, and the average imaging cost per treated Code Stroke patient, defined as a case resulting in IVT or EVT treatment. Results From 2020 to 2025, annual Code Stroke activations and associated imaging expenditures rose nearly three-fold from 133 to 298 cases, driven by an approximately 800% increase in CTA head and neck utilization. Annual imaging expenditure increased from $70,616 to $259,290.Despite the increasing number of Code Stroke activations, the number of patients receiving IVT or EVT has remained stable (19 cases in 2020 vs 16 in 2025), with treatment yield declining from 14.3% to 5.4%. Consequently, the proportion of Code Strokes resulting in reperfusion decreased over time, with the average cost per treated patient increasing over 4-fold from $3717 to $16,205. Conclusion Our findings describe a disproportionate rise in imaging utilization and cost relative to the number of treated Code Strokes. These findings characterize longitudinal trends in imaging utilization and resource use within contemporary stroke evaluation pathways.
Purpose To evaluate the effects of supraclavicular radiotherapy on carotid arteries using Doppler ultrasonography by comparing irradiated and non-irradiated sides within the same patients. Methods This retrospective observational study included 50 female patients who had previously undergone supraclavicular radiotherapy for breast cancer. All participants were evaluated using carotid Doppler ultrasonography. The mean interval between completion of radiotherapy and ultrasonographic evaluation was 6.1 ± 2.7 years. Carotid artery diameters, intima–media thickness (IMT), and Doppler parameters (peak systolic velocity [PSV] and end-diastolic velocity [EDV]) were measured on both irradiated (RT+) and non-irradiated (RT−) sides. Correlation analyses between time since radiotherapy and vascular parameters were performed. Results IMT values were significantly higher on the irradiated side for both the common carotid artery (CCA) and internal carotid artery (ICA) (p = 0.001 for both). CCA diameters were significantly increased on the irradiated side, whereas ICA diameters showed no significant differences. Doppler parameters (PSV and EDV) did not differ significantly between irradiated and non-irradiated sides (p > 0.05 for all). No significant correlation was observed between time since radiotherapy and IMT values. Conclusion Supraclavicular radiotherapy is associated with significant structural changes in carotid arteries, particularly increased IMT, while hemodynamic flow parameters remain preserved. IMT measurement may serve as a useful non-invasive tool for early detection of radiation-induced vascular injury.
Purpose To evaluate radiologists' early adoption, perceptions, and radiologist-reported clinical impact of breast cancer detection artificial intelligence (AI) tools in breast imaging practices. Methods An online survey was distributed to members of the Society of Breast Imaging to assess perceived clinical impact of FDA-cleared AI tools for mammographic breast cancer detection. Respondents were categorized as AI users or non-users. Responses were summarized descriptively, and differences were analyzed using Fisher's exact test. Results A total of 215 radiologists responded. Of these, 47% had implemented breast cancer detection AI tools (AI users), 11.2% are planning to implement them, and 41.8% had not implemented diagnostic AI tools (non-AI users). Among non-AI users, the most reported barrier to adoption was the implementation cost (53.3%). The most frequently perceived appropriate use of diagnostic AI tools was as a second reader. Non-AI users more often anticipated reductions in recall rates (59.3%) compared with AI users reporting reductions (34.7%; p = 0.003). A similar pattern was observed for biopsy rates (36.4% vs 9.1%; p < 0.001). Non-AI users also more frequently anticipated reduced burnout than AI users (56.0% vs 29.4%; p < 0.001). Perceptions of patient outcomes were similar between groups (p = 0.87). Conclusion Although early adoption of breast cancer detection AI tools is notable, only a minority of AI users reported meaningful clinical benefit across measured outcomes, falling short of the benefits anticipated by non-AI users. Continued validation and refinement of AI tools is needed to ensure meaningful clinical impact.
BACKGROUND:Post-tuberculosis lung disease affects 30-50% of TB survivors despite successful treatment, yet systematic post-TB screening and surveillance remain absent from most national TB programs. Radiographic heterogeneity predicts functional impairment, but relationships between radiographic phenotypes and functional outcomes remain incompletely characterized in high-burden TB settings. METHODS:This prospective cross-sectional study enrolled 500 TB survivors (6-60 months post-treatment). PRIMARY OUTCOME:radiographic phenotype classification (minimal-change/moderate-complexity/complex) and associations with pulmonary function impairment (FEV₁ <80% predicted). Chest radiography assessed cavitation, bronchiectasis, parenchymal opacification, architectural distortion via standardized morphometric scoring. Spirometry performed per ATS/ERS 2019 standards. Inflammatory biomarkers (hs-CRP, fibrinogen) measured. Radiographic inter-rater reliability assessed via ICC. Logistic regression identified independent predictors. RESULTS:Of 625 enrolled, 500 (80%) completed assessment. Radiographic phenotypes: minimal-change 31.2%, moderate-complexity 42.8%, complex 26.0%. FEV₁ % predicted: minimal-change 89.2 ± 10.4%, moderate 75.8 ± 14.2%, complex 46.2 ± 16.8% (p < 0.001). FEV₁ <80% predicted: minimal 11.5%, moderate 56.1%, complex 98.5% (p < 0.001). Radiographic inter-rater reliability excellent (ICC 0.84; 95% CI 0.81-0.87). Complex phenotype associated with FEV₁ <80% (aOR 28.48; 95% CI 10.12-80.04). Elevated hs-CRP present in 24.4% minimal-change versus 76.9% complex (p < 0.001). Inflammatory biomarkers mediated 55% of phenotype effect on function. CONCLUSIONS:Chest radiographic phenotypes robustly predict pulmonary function impairment in post-TB disease. Complex phenotypes show severe obstruction (98.5%) and elevated inflammation. Simple, protocol-based morphometric scoring (3.8 min per image, ICC 0.84 between trained readers) is feasible for resource-limited settings. Radiographic phenotyping enables risk stratification in TB survivors.
PURPOSE:Spinal vascular malformations (SVMs) are rare lesions requiring accurate imaging. Two-dimensional digital subtraction angiography (2D DSA) remains the diagnostic gold standard but is invasive and limited in three-dimensional localization. Magnetic resonance angiography (MRA), computed tomography angiography (CTA), and three-dimensional rotational angiography (3D RA) represent alternative techniques. This study aimed to assess the diagnostic accuracy of MRA, CTA, and 3D RA versus 2D DSA in SVMs and to provide an overview of misdiagnoses and advances in image processing. METHODS:A systematic review and bivariate meta-analysis of diagnostic studies was conducted in accordance with PRISMA and following a predefined protocol. Pooled sensitivity and specificity were calculated, and risk of bias was assessed using the QUADAS-2 tool. RESULTS:MRA, CTA and 3D RA demonstrated high diagnostic sensitivity (86.6%, 87.7% and 86.7%, respectively) and variable specificity (64.2%, 73.1% and 85%, respectively). Delayed or missed diagnoses were frequently related to incomplete angiographic coverage or technical limitations of 2D DSA. Image fusion and advanced post-processing techniques improved diagnostic confidence. CONCLUSION:Alternative modalities show substantial potential as adjuncts to 2D DSA, particularly when combined with advanced post-processing. Standardized imaging protocols and prospective studies are needed to define their role in clinical practice.
OBJECTIVES:To compare the diagnostic accuracy of contrast-enhanced mammography (CEM) with digital mammography (DM), digital breast tomosynthesis (DBT) and ultrasonography (USG) for breast cancer detection in dense breasts. METHODS:This prospective study was conducted between January 2023 and 2025 including DM with ACR- C or D category parenchymal density. Imaging findings were categorized using the ACR BI-RADS assessment where BI-RADS scores 4 and 5 were considered suspicious for malignancy. Histopathology was reference standard for such lesions and imaging follow-up for categories 1 to 3. Sensitivity, specificity, positive and negative predictive value (NPV), likelihood & odds ratios, ROC analysis, and kappa statistics were calculated. RESULTS:In 100 females (mean age, 45 ± 7.2 years), CEM demonstrated the highest sensitivity (97.9%) and NPV (97.7%), outperforming DM as well as DBT (p < 0.001). Subgroup analysis was performed for diagnostic (n = 56) and surveillance cohort (n = 44). In the former, CEM achieved 100% sensitivity and identified additional multifocal, multicentric, or contralateral disease in 12 patients. It also demonstrated higher specificity (94.3%) and strongest association with malignancy in surveillance group. Agreement with histopathology was highest for CEM (κ 0.80). CONCLUSION:CEM demonstrates superior diagnostic accuracy than conventional mammographic techniques in dense breasts, with excellent sensitivity, NPV and agreement with histopathology. ADVANCES IN KNOWLEDGE:CEM improves cancer detection and disease extent assessment in dense breasts, enabling confident exclusion of malignancy and more accurate staging. Its high diagnostic performance supports use as practical adjunct, particularly in resource-limited settings.
PURPOSE:To characterize the spectrum, severity and temporal pattern of adverse events reported for implantable breast markers, and to compare reported-problem profiles across device subtypes, localization technologies and clip brand families. METHODS:We analyzed the U.S. FDA Manufacturer and User Facility Device Experience (MAUDE) database, a national passive post-market surveillance system. Reports under product code NEU (implantable soft-tissue marker) received during 2010-2025 were retrieved via openFDA, deduplicated and restricted to breast markers after excluding non-breast devices. Device- and patient-problem codes were compared across device subtypes, localization technology classes and clip brand families (chi-square; Fisher exact test). MAUDE has no usage denominator, so incidence was not estimated. RESULTS:Of 1503 reports, 51.8% involved biopsy-site clips, 39.9% BioZorb and 8.3% localization devices; 65.0% were injuries. Reporting rose sharply in 2023-2025, driven almost entirely by BioZorb (81% postdating the February 2024 FDA communication). BioZorb reports were far more likely than other markers to describe pain (odds ratio [OR] 52.5), infection (OR 27.6), severe local tissue effects (OR 69.3) and deformity (OR 275). Profiles differed within subtypes: radar reflectors were malfunction-predominant with no migration reports, whereas RFID tags accounted for most localization-device migration; among clips, HydroMark and CeleroMark/SMark reports had the highest proportions of allergic or cutaneous complaints. CONCLUSION:Complications were reported for all categories of implantable breast markers, with distinct profiles by device technology and brand family. The recent reporting surge reflects recall-driven reporting of BioZorb complications.
PURPOSE:Gender disparities persist in academic radiology, particularly in leadership roles. However, gender representation across radiology journal editorial boards has not been comprehensively evaluated. This study evaluated gender representation across editorial roles in radiology journals and to identify the factors associated with gender disparity. METHODS:In this cross-sectional retrospective study, radiology journals listed in the Web of Science Core Collection were screened between November 2025 and February 2026. Editors were categorized by role, and gender was determined using publicly available information. Journal characteristics, including indexing status, quartile, geographic origin, subspecialty, publishing model, impact factor, and publication duration were recorded. Gender distributions were compared using chi-square tests. Factors associated with gender disparity were evaluated using logistic regression. RESULTS:A total of 166 journals comprising 9506 editors were included. Women were underrepresented across all editorial roles, accounting for 21.0% of editors-in-chief and 26.6% of editorial board members. Female representation differed significantly according to journal characteristics (all p < 0.05). In multivariable analysis, emerging sources citation indexing (OR:0.14; 95% CI, 0.04-0.48, p = 0.002), society-association journal status (OR:0.27; 95% CI, 0.10-0.74, p = 0.011), and longer publication duration (OR:0.30; 95% CI, 0.10-0.89, p = 0.031) were associated with lower odds of gender disparity. Absence of a female editor-in-chief was independently associated with higher odds of gender disparity (OR:4.66; 95% CI, 1.87-11.63, p = 0.001). CONCLUSION:Women remain underrepresented in editorial roles in radiology journals, with variation associated with journal characteristics and leadership composition.
PURPOSE:To preliminarily evaluate whether routine knee MRI demonstrates reproducible patterns of combined patellar cartilage and subchondral bone involvement using an exploratory three-tier patella-focused osteochondral MRI framework. METHODS:This retrospective single-center lesion-based study included 200 knee MRI examinations performed between January 1, 2024 and January 1, 2026. Patellar cartilage damage was graded as Amax (0-4), and subchondral bone abnormalities were graded as Bmax (0-3). Examinations were categorized as Tier 1 low-burden involvement, Tier 2 discordant involvement, or Tier 3 concordant advanced involvement. Field-strength subgroup analyses were performed to assess potential differences between 1.5-T and 3-T examinations. RESULTS:Interobserver agreement was high for Amax (weighted kappa = 0.82) and Bmax (weighted kappa = 0.85), and excellent for the composite score (ICC = 0.91). Cartilage and bone grades were positively correlated (Spearman's rho = 0.637, p < 0.001). Tier 1, Tier 2, and Tier 3 patterns were observed in 43.5%, 23.5%, and 33.0% of examinations, respectively. Within Tier 2, cartilage-predominant and bone-predominant patterns accounted for 38/47 (80.9%) and 9/47 (19.1%) cases, respectively. Field-strength analysis showed a difference in detailed Amax grade distribution (p = 0.021), whereas high-grade cartilage involvement, Bmax distribution, framework category, and composite score did not differ significantly between 1.5-T and 3-T examinations. CONCLUSION:Routine knee MRI may reveal low-burden, discordant, and concordant advanced patterns of patellar osteochondral involvement. The proposed framework provides an exploratory structural description of combined cartilage-bone involvement rather than a validated staging system or a proven replacement for conventional grading.
PURPOSE:To systematically evaluate the impact of MRI software updates on apparent diffusion coefficient (ADC) values, assess potential left-right asymmetry in bilateral breast ADC values, and explore standardization techniques to enhance diagnostic reliability in lesion classification. METHODS:A total of 420 patients (mean age, 54.2 ± 14.2 years) who underwent 3 T breast MRI between 2021 and 2025 were retrospectively analyzed. The ADC values in fibroglandular tissue (FGT) and pectoralis major muscle (PMM) were compared before and after a software update using an unpaired t-test. Left-right asymmetries were assessed using paired and unpaired t-tests. Differences in lesion ADC values between malignancies were analyzed. The diagnostic performance of three region of interest (ROI) methods-whole-lesion mean (WL-mean), small-ROI mean (SR-mean), and small-ROI minimum (SR-min)-was evaluated using receiver operating characteristic analysis. The effect of standardization using ratios or differences relative to FGT and PMM on diagnostic accuracy was also examined. RESULTS:The ADC values were affected by the software update, and notable left-right asymmetries were observed in both FGT and PMM. Malignancies had significantly lower ADC values than benign lesions. Among the ROI methods, SR-min yielded the highest diagnostic accuracy, although diagnostic performance was lower in non-mass lesions than in mass lesions across the ROI methods. Standardization using ratios or differences relative to FGT and PMM did not improve diagnostic performance. CONCLUSIONS:ADC measurements are sensitive to technical variables and exhibit unexpected laterality effects, thereby complicating standardization. While general thresholds may guide interpretation, establishing universally standardized ADC values remains challenging.
OBJECTIVE:To evaluate patient perspectives on breast cancer, risk assessment, and screening tools. METHODS:This IRB-approved HIPAA-compliant study included an anonymous 12-question survey administered to female patients at 4 medical clinics (2 radiology, 1 internal medicine and 1 gynecology) in Northern California during May-June 2024. Results were compared using Fisher exact tests between groups. A p-value of <0.05 was considered significant. RESULTS:Surveys were completed by 319 patients at radiology (160), gynecology (80) and internal medicine (79) clinics. On a scale of 1 (not concerned) to 5 (extremely concerned), 34.7% (111/319) reported 4 or 5 out of 5 concern for developing breast cancer. Among respondents not aware of their lifetime risk, 77.2% (85/110) were interested in obtaining a risk assessment. Most respondents (80.5% (257/319)) reported they would be comfortable with breast cancer risk assessment derived from their screening mammogram by AI-based methods, compared to 31.0% (99/319) using personal/family history obtained from a survey. Patients surveyed at gynecology and radiology offices preferred obtaining their risk results from their gynecologist (82.5% (66/80)) (p ≤ 0.001), (57.3% (90/157)) (p = 0.006), and patients surveyed at the internal medicine clinic (63.3% (50/79) (p = 0.081) preferred their PCP. If found to be at increased risk, 90.3% (288/319) of patients reported interest in additional screening tests. Most patients at increased breast cancer risk (72% (111/154)) reported interest in genetic testing, compared with 50% (84/165) of patients not at increased breast cancer risk (p = 0.001). CONCLUSION:Patients demonstrated a strong interest in AI-based risk assessment using their screening mammogram, and those at higher risk expressed interest in additional breast cancer screening tools and genetic testing.
