Radiology Residents Research Articles

Developing a Virtual Radiology Elective for Female and Underrepresented in Medicine Applicants.

Developing a Virtual Radiology Elective for Female and Underrepresented in Medicine Applicants.

Incidence and predictors of discrepancies in radiology resident interpretations of coronary CT in the emergency department

BackgroundDiscrepancies between preliminary reports by on-call radiology residents and final reports of coronary computed tomography angiography (CCTA) in the emergency department (ED) have not been thoroughly investigated.MethodsWe conducted a retrospective quality assurance analysis of CCTA examinations performed during off-hours in a level-1 ED at a tertiary teaching hospital between March 2020 and April 2022. Discrepancies in identifying significant coronary artery disease (≥ 50% stenosis) between preliminary reports by on-call residents and final reports by board-certified cardiac radiologists were evaluated.ResultsAmong the 766 patient visits (median age, 59 years [interquartile range, 47–70]; 415 men), 82 cases (10.7%) showed discrepancies. Univariable logistic regression analyses identified HEART score, day of ED visit, ED crowding index, and coronary artery calcium (CAC) score as significant factors associated with discrepancies. Multivariable analysis revealed that an ED crowding index < 40 (adjusted odds ratio = 2.06; P = 0.005), and positive CAC scores were independently associated with increased discrepancies (adjusted odds ratio = 4.56 for scores > 0 and ≤ 100, P < 0.001; 4.79 for scores > 100 and ≤ 400, P < 0.001; 3.69 for scores > 400, P = 0.002). The rate of unnecessary invasive coronary angiography was significantly higher in the discrepancy group (80.0%, 12 of 15) compared to the agreement group (14.4%, 16 of 111) (P < 0.05).ConclusionsA substantial discrepancy rate was observed between preliminary and final CCTA interpretations in the ED. A lower ED crowding index and positive CAC scores were independently associated with an increased risk of discrepancies.

Play in the reading room: Utilizing soft modeling compound to teach musculoskeletal anatomy and pathology.

Play in the reading room: Utilizing soft modeling compound to teach musculoskeletal anatomy and pathology.

Utilization of artificial intelligence-based pulmonary nodule target reconstruction software in clinical practice education for standardized training residents

BackgroundThe precise evaluation of pulmonary nodules via computed tomography (CT) is pivotal in clinical decision-making and patient prognosis. Artificial intelligence (AI)-assisted target reconstruction technology for pulmonary nodules can more clearly display the target nodule and its relationship with surrounding structures, aiding radiologists in higher diagnostic accuracy. Nevertheless, few studies have examined the impact of such AI-based software on the education and training of residents. This study aims to investigate the role of AI-assisted pulmonary nodule target reconstruction software in enhancing the diagnostic capabilities of residents from different specialties and to explore differences between three different learning modes, thereby preliminarily assessing the significance of AI technology in clinical training of medical imaging for residents.MethodsSeventy-five standardized training residents from various specialties, including 32 radiology and 43 non-radiology residents, participated in rotations in our radiology department between August 2020 and September 2023. Following a four-week period of training and learning with AI-assisted pulmonary nodule target reconstruction software and the traditional picture archiving and communication system (PACS), the diagnostic capabilities of both radiology and non-radiology residents in evaluating pulmonary nodule cases were assessed. Additionally, the differences in their ability to assess and diagnose pulmonary nodules under three distinct learning modes assisted by AI software (full-application, cross-application, and interval-application) were analyzed.ResultsAfter four weeks of training, the diagnostic accuracy of radiology residents for five test pulmonary nodule cases ranged from 96.88 to 100%, outperforming non-radiology residents, whose accuracy ranged from 67.44 to 86.04%. Among the 54 residents trained under three predefined learning modes, significant differences were found in pulmonary nodule assessment scores. Pairwise comparisons using the Tukey-Kramer test revealed that the full-application group scored lower compared to both the cross-application (p = 0.002) and interval-application (p = 0.004) groups, with the latter two demonstrating superior performance.ConclusionAI-assisted target reconstruction and assessment software for pulmonary nodules is found to be valuable in medical imaging education and training. A hybrid learning approach that integrates AI software with traditional PACS may be more effective in enhancing the pulmonary nodule assessment and diagnostic capabilities of residents.

Leveraging ChatGPT for Enhancing Learning in Radiology Resident Education.

Leveraging ChatGPT for Enhancing Learning in Radiology Resident Education.

Musculoskeletal imaging fellowship program directors: A study of educational paths, and scholarly engagement in the United States

Musculoskeletal imaging fellowship program directors: A study of educational paths, and scholarly engagement in the United States

The current state of radiology residency lectures: Requirements, institutional approach, challenges, and solutions.

The current state of radiology residency lectures: Requirements, institutional approach, challenges, and solutions.

Development and validation of AI-based automatic segmentation and measurement of thymus on chest CT scans

BackgroundDue to the complex anatomical structure and dynamic involution process of the thymus, segmentation and evaluation of the thymus in medical imaging present significant challenges. The aim of this study is to develop a deep-learning tool “Thy-uNET” for automatic segmentation and measurement of the thymus or thymic region on chest CT imaging, and to validate its performance with multicenter data.Materials and methodsUtilizing the segmentation and measurement results from two experts, training of Thy-uNET was conducted on training cohort (n = 500). The segmented regions include thymus or thymic region, and 7 features of the thymic region were measured. The automatic segmentation performance was assessed using Dice and Intersection over Union (IOU) on CT data from three test cohorts (n = 286). Spearman correlation analysis and intraclass correlation coefficient (ICC) were used to evaluate the correlation and reliability of the automatic measurement results. Six radiologists with varying levels of experience were invited to participate in a reader study to assess the measurement performance of Thy-uNET and its ability to assist doctors.ResultsThy-uNET demonstrated consistent segmentation performance across different subgroups, with Dice = 0.83 in the internal test set, and Dice = 0.82 in the external test sets. For automatic measurement of thymic features, Thy-uNET achieved high correlation coefficients and ICC for key measurements (R = 0.829 and ICC = 0.841 for CT attenuation measurement). Its performance was comparable to that of radiology residents and junior radiologists, with significantly shorter measurement time. Providing Thy-uNET measurements to readers reduced their measurement time and improved residents’ performance in some thymic feature measurements.ConclusionThy-uNET can provide reliable automatic segmentation and automatic measurement information of the thymus or thymic region on routine CT, reducing time costs and improving the consistency of evaluations.

Interpreting Chest X-ray with ChatGPT: Can It Serve as a Tool for Justifying Computed Tomography?

Objective: The aim of this study was to test the success of ChatGPT-4 in evaluating chest radiographs and detecting abnormal findings, and then to demonstrate its utility in computed tomography (CT) justification. Methods: This study included 59 patients (20 patients in the first phase, and 39 patients in the second phase) from a publicly available chest X-ray dataset. X-rays were evaluated by an experienced chest radiologist (as gold standard), two radiology residents, and ChatGPT, first as normal-abnormal and then whether CT was needed if abnormal. Finally, the ChatGPT and two radiology residents' decisions were compared with the gold standard decision of the expert radiologist to obtain an accuracy value. Results: The accuracy of Resident 1, Resident 2, and ChatGPT for normal-abnormal labeling was 76.27%, 93.22%, and 76.27%, respectively, for a total of 59 patients. The accuracy of Resident 1, Resident 2, and ChatGPT for CT necessity was 67.80%, 72.88%, and 66.10%, respectively. The expert radiologist determined that CT was not necessary in 30 patients. Of these 30 patients, Resident 1, Resident 2, and ChatGPT answered incorrectly in 14, 12, and 15 patients, respectively. There is no statistically significant difference between the responses of Resident 1, Resident 2, and ChatGPT for CT necessity (Chi-square, p=0.731). Conclusion: The results of this study show that ChatGPT-4 is promising for chest X-ray interpretation and justification of CT scans. However, large language models such as ChatGPT, which still have major limitations, should be trained with a much larger number of radiology images.

Automated breast ultrasound features associated with diagnostic performance of Multiview convolutional neural network according to radiologists' experience.

To investigate automated breast ultrasound (ABUS) features affecting the use of Multiview convolutional neural network (CNN) for breast lesions according to radiologists' experience. A total of 656 breast lesions (152 malignant and 504 benign lesions) were included and reviewed by six radiologists for background echotexture, glandular tissue component (GTC), and lesion type and size without as well as with Multiview CNN. The sensitivity, specificity, and the area under the receiver operating curve (AUC) for ABUS features were compared between two sessions according to radiologists' experience. Radiology residents showed significant AUC improvement with the Multiview CNN for mass (0.81 to 0.91, P=0.003) and non-mass lesions (0.56 to 0.90, P=0.007), all background echotextures (homogeneous-fat: 0.84 to 0.94, P=0.04; homogeneous-fibroglandular: 0.85 to 0.93, P=0.01; heterogeneous: 0.68 to 0.88, P=0.002), all GTC levels (minimal: 0.86 to 0.93, P=0.001; mild: 0.82 to 0.94, P=0.003; moderate: 0.75 to 0.88, P=0.01; marked: 0.68 to 0.89, P<0.001), and lesions ≤10mm (≤5 mm: 0.69 to 0.86, P<0.001; 6-10 mm: 0.83 to 0.92, P<0.001). Breast specialists showed significant AUC improvement with the Multiview CNN in heterogeneous echotexture (0.90 to 0.95, P=0.03), marked GTC (0.88 to 0.95, P<0.001), and lesions ≤10mm (≤5 mm: 0.89 to 0.93, P=0.02; 6-10 mm: 0.95 to 0.98, P=0.01). With the Multiview CNN, the performance of ABUS in radiology residents was improved regardless of lesion type, background echotexture, or GTC. For breast lesions smaller than 10 mm, both radiology residents and breast specialists showed better performance of ABUS.

CONCORDANCE CONCORDANCE BETWEEN JUNIOR RESIDENTS AND CONSULTANT RADIOLOGISTS IN REPORTING PNEUMOPERITONEUM ON PLAIN RADIOGRAPHS

Background: Pneumoperitoneum, the presence of free intraperitoneal air, is a critical radiological finding often indicative of gastrointestinal perforation and requires immediate intervention. Early detection using plain radiographs is essential, especially in resource-limited settings where advanced imaging may not be readily available. However, interpretation accuracy may vary with clinical experience, particularly during on-call hours when junior residents are primarily responsible for initial assessments. Establishing the reliability of resident interpretations is vital to improving diagnostic workflows and patient outcomes. Objective: To assess the level of diagnostic concordance between junior radiology residents and consultant radiologists in identifying pneumoperitoneum on plain radiographs and to analyze variations across demographic and clinical subgroups. Methods: A cross-sectional study was conducted over six months (December 9, 2021, to June 8, 2022) at the Department of Diagnostic Radiology, Aga Khan University Hospital, Karachi. A total of 100 radiographs were prospectively analyzed. First- and second-year FCPS-II radiology residents independently assessed anonymized plain radiographs for signs of pneumoperitoneum, categorizing each as negative or requiring urgent attention. These preliminary evaluations were then compared with final consultant reports. Inter-observer agreement was quantified using Cohen’s Kappa statistic, with stratification based on age, gender, patient location, radiographic technique, and residency year. Results: The mean age of the patients was 38.09 ± 17.48 years, with 61.0% male and 39.0% female participants. Junior residents identified pneumoperitoneum in 26 cases, while consultant radiologists confirmed 74 cases. Diagnostic concordance was observed in 82 out of 100 cases. The Kappa coefficient was 0.520 (95% CI: 0.327–0.714, p &lt; 0.001), indicating moderate agreement. Substantial agreement was found among patients aged &lt;60 years (κ = 0.684), females (κ = 0.692), and ICU/outpatient settings (κ = 0.750, κ = 0.765). Decubitus radiographs demonstrated perfect agreement (κ = 1.000), while supine views showed lower agreement (κ = 0.298). Conclusion: This study demonstrates moderate yet statistically significant diagnostic agreement between junior residents and consultants in identifying pneumoperitoneum. Variations in concordance across subgroups highlight the need for enhanced supervision, feedback mechanisms, and targeted radiographic interpretation training to improve diagnostic reliability among junior radiologists.

Tail Lights

I hope everyone has enjoyed the wonderful Spring weather, this academic year and is now looking forward to the Summer holidays! (Not too long to go now till you too are done, MB4!) What a busy few months it has been for you. MedChir Revue in May was of course a highlight in the calendar. A fantastic array of talent was on display – from high culture opera to artistic sports people. Wonderful to see the (it-must-be-a-record-breaking-number-of) judges unanimously award the top prize to the terrific Otago Street Boys (final years William Heeley, Rueben Heaton and William Strachan). Bravo and totally well deserved! It is great to see some of your audits being submitted as Visual Abstracts. Keep them coming! Remember: submissions accepted for publication receive £20 and is legitimate CV fodder (citable DOI)! Students presenting their work at conferences is always impressive. Everyone was blown away by one of our 3rd year students competing against Radiology Residents at the Scottish Radiological Society Spring Meeting for the SRS Dr Sarah Jenkins QI Prize. It was a close run thing and she was just pipped to the post. You can read Shraddha’s visual abstract in this issue of Surgo and the other works here: https://www.radiology.co.uk/spring-2025-qi-presentations. Graduation is in a couple of weeks while 335 MB5’s are heading to Crieff for their Graduation Ball. Epsilon 2025 is a significant cohort for many reasons. As a 25-year old Epsilonian myself, I am extra proud of this group of graduands. Not only are they the first cohort to officially sit the GMC Medical Licencing Assessment, they also did this after starting their medical school journey during the COVID pandemic. What a fantastic achievement! Super well done everyone! YOU ARE THE OG! Remember you have resilience in spades – the world is your oyster: go make your mark! Dr Christine McAlpine is the subject of this issue’s interview. Be inspired by her incredible journey from being the first woman president of MedChir Society to Consultant Stroke Physician at Glasgow Royal Infirmary today. Until the next Tail Lights, have a fabulous summer and see you all in the Autumn! Remember to send in your holiday pictures as Surgo Vision or turn your audits into Visual Abstracts!

Radiology resident education during the COVID-19 pandemic in the United States

The COVID-19 pandemic presented unprecedented challenges to healthcare systems, including radiology residency programs. This study aims to examine the impact of COVID-19 on radiology residency education and identify interventions implemented for future unplanned disruptions to physician training. Data collection occurred between March to April 2022 through a survey distributed to 30 radiology residency program directors from diverse geographic regions, hospital types, and practice settings. Data was collected on program characteristics, COVID-19 impact, changes in scheduling and teaching methods, and perceived effects on resident competence and well-being. All surveyed programs implemented changes to address resident teaching to accommodate social distancing. Most programs (86.7%) offered remote work/study options. A majority (66.7%) implemented alternating resident schedules. Virtual conferences and virtual view-box teaching were identified as the most utilized interventions during social distancing requirements. The majority (76.1%) of programs reported worsened resident education during the pandemic, with first-year residents the most adversely affected group. Decreased competence was noted in 40% of first-year and 36.7% of second-year residents compared to pre-pandemic cohorts. Additionally, a significant portion (73.3%) of program directors reported negative impacts on resident well-being. The COVID-19 pandemic significantly disrupted radiology residency despite mitigation efforts. While virtual teaching methods provided necessary alternatives during the pandemic, they could not fully replace traditional in-person education, as evidence by widespread reports of worsened educational outcomes. Recommendations for future preparedness include prioritizing early deployment of remote workstations, incorporating alternative teaching methods, providing increased on-site instruction for junior residents, and enhancing mental health support. These lessons can inform strategies to better prepare residency programs for future challenges and ensure the continued production of competent, resilient radiologists.

Efficacy of a large language model in classifying branch-duct intraductal papillary mucinous neoplasms.

Appropriate categorization based on magnetic resonance imaging (MRI) findings is important for managing intraductal papillary mucinous neoplasms (IPMNs). In this study, a large language model (LLM) that classifies IPMNs based on MRI findings was developed, and its performance was compared with that of less experienced human readers. The medical image management and processing systems of our hospital were searched to identify MRI reports of branch-duct IPMNs (BD-IPMNs). They were assigned to the training, validation, and testing datasets in chronological order. The model was trained on the training dataset, and the best-performing model on the validation dataset was evaluated on the test dataset. Furthermore, two radiology residents (Readers 1 and 2) and an intern (Reader 3) manually sorted the reports in the test dataset. The accuracy, sensitivity, and time required for categorizing were compared between the model and readers. The accuracy of the fine-tuned LLM for the test dataset was 0.966, which was comparable to that of Readers 1 and 2 (0.931-0.972) and significantly better than that of Reader 3 (0.907). The fine-tuned LLM had an area under the receiver operating characteristic curve of 0.982 for the classification of cyst diameter ≥ 10mm, which was significantly superior to that of Reader 3 (0.944). Furthermore, the fine-tuned LLM (25s) completed the test dataset faster than the readers (1,887-2,646s). The fine-tuned LLM classified BD-IPMNs based on MRI findings with comparable performance to that of radiology residents and significantly reduced the time required.

RadGPT: A system based on a large language model that generates sets of patient-centered materials to explain radiology report information.

RadGPT: A system based on a large language model that generates sets of patient-centered materials to explain radiology report information.

Curriculum check, 2025-equipping radiology residents for AI challenges of tomorrow.

The exponential rise in the artificial intelligence (AI) tools for medical imaging is profoundly impacting the practice of radiology. With over 1000 FDA-cleared AI algorithms now approved for clinical use-many of them designed for radiologic tasks-the responsibility lies with training institutions to ensure that radiology residents are equipped not only to use AI systems, but to critically evaluate, monitor, respond to their output in a safe, ethical manner. This review proposes a comprehensive framework to integrate AI into radiology residency curricula, targeting both essential competencies required of all residents, optional advanced skills for those interested in research or AI development. Core educational strategies include structured didactic instruction, hands-on lab exposure to commercial AI tools, case-based discussions, simulation-based clinical pathways, teaching residents how to interpret model cards, regulatory documentation. Clinical examples such as stroke triage, Urinary tract calculi detection, AI-CAD in mammography, false-positive detection are used to anchor theory in practice. The article also addresses critical domains of AI governance: model transparency, ethical dilemmas, algorithmic bias, the role of residents in human-in-the-loop oversight systems. It outlines mentorship, faculty development strategies to build institutional readiness, proposes a roadmap to future-proof radiology education. This includes exposure to foundation models, vision-language systems, multi-agent workflows, global best practices in post-deployment AI monitoring. This pragmatic framework aims to serve as a guide for residency programs adapting to the next era of radiology practice.

Entrustable professional activities in Finnish radiology training: a national survey

ObjectivesThis study assessed the practical implementation, experiences, and attitudes toward entrustable professional activities (EPAs) in radiology training across Finland.MethodsA nationwide, anonymous online survey targeted radiology residents, recently graduated specialists (within 3 years), and instructor specialists. Distributed to all Finnish hospitals involved in radiology training, the survey evaluated EPA completion rates, perceived value, and future development needs. Responses were analyzed to identify trends and differences across groups.ResultsOf 150 respondents (42% residents, 43% instructors, and 14% recent graduates), 65% were from university hospitals. Among residents and recent graduates, 37% had completed EPA assessments, with 87% valuing the feedback received and 73% finding EPAs effective for competency assessment. Overall, 64% considered EPAs well-suited to radiology. Residents showed higher completion rates (43%) than recent graduates (19%), with fourth- and fifth-year residents more engaged (69% vs. 15%). Instructors, while supportive (67% viewed EPAs as meaningful), emphasized a need for more training (54% vs. 49% of residents).ConclusionMost Finnish radiology respondents considered EPAs well-suited for training. Residents and recent graduates who completed EPAs greatly valued the feedback and found them effective for assessing competencies, with residents participating more actively than recent graduates. Instructors’ desire for better guidance suggests a priority for enhanced support and education. These findings endorse EPA integration and inform refinements in national and European radiology curricula.Critical relevance statementFinnish radiologists and residents strongly support EPAs in radiology training, valuing their feedback and competency assessment, though instructors seek enhanced guidance.Key PointsFinnish radiology residents and specialists reported positive experiences and strong support for entrustable professional activity (EPAs).Finland’s mandatory, nationally coordinated EPA framework contrasts with subspecialty-focused models elsewhere.Instructors seek more EPA training, signaling a need for enhanced education to sustain their engagement as adoption grows.Graphical

Boosting LLM-assisted diagnosis: 10-minute LLM tutorial elevates radiology residents' performance in brain MRI interpretation.

Boosting LLM-assisted diagnosis: 10-minute LLM tutorial elevates radiology residents' performance in brain MRI interpretation.

Assessment of a Task Trainer and Targeting Tasks for Ultrasound-Guided Invasive Procedures in Radiology Residents.

Assessment of a Task Trainer and Targeting Tasks for Ultrasound-Guided Invasive Procedures in Radiology Residents.

Improving patient safety education for radiology residents: Using a quality improvement approach.

Improving patient safety education for radiology residents: Using a quality improvement approach.