Radiology
Recent Submissions
-
A Prospective Analysis of Screen-Detected Cancers Recalled and Not Recalled by Artificial Intelligence.Objective: The use of artificial intelligence has potential in assisting many aspects of imaging interpretation. We undertook a prospective service evaluation from March to October 2022 of Mammography Intelligent Assessment (MIA) operating "silently" within our Breast Screening Service, with a view to establishing its performance in the local population and setting. This evaluation addressed the performance of standalone MIA vs conventional double human reading of mammograms. Methods: MIA analyzed 8779 screening events over an 8-month period. The MIA outcome did not influence the decisions made on the clinical pathway. Cases were reviewed approximately 6 weeks after the screen reading decision when human reading and/or MIA indicated a recall. Results: There were 146 women with positive concordance between human reading and MIA (human reader and MIA recalled) in whom 58 breast cancers were detected. There were 270 women with negative discordance (MIA no recall, human reader recall) for whom 19 breast cancers and 1 breast lymphoma were detected, with 1 cancer being an incidental finding at assessment. Six hundred and four women had positive discordance (MIA recall, human reader no recall) in whom 2 breast cancers were detected at review. The breast cancers demonstrated a wide spectrum of mammographic features, sites, sizes, and pathologies, with no statistically significant difference in features between the negative discordant and positive concordant cases. Conclusion: Of 79 breast cancers identified by human readers, 18 were not identified by MIA, and these had no specific features or site to suggest a systematic error for MIA analysis of 2D screening mammograms. Keywords: National Health Service Breast Screening Programme; artificial intelligence; breast cancer screening; digital mammography; double reading
-
Development and validation of open-source deep neural networks for comprehensive chest x-ray reading: a retrospective, multicentre studyBackground: Artificial intelligence (AI) systems for automated chest x-ray interpretation hold promise for standardising reporting and reducing delays in health systems with shortages of trained radiologists. Yet, there are few freely accessible AI systems trained on large datasets for practitioners to use with their own data with a view to accelerating clinical deployment of AI systems in radiology. We aimed to contribute an AI system for comprehensive chest x-ray abnormality detection. Methods: In this retrospective cohort study, we developed open-source neural networks, X-Raydar and X-Raydar-NLP, for classifying common chest x-ray findings from images and their free-text reports. Our networks were developed using data from six UK hospitals from three National Health Service (NHS) Trusts (University Hospitals Coventry and Warwickshire NHS Trust, University Hospitals Birmingham NHS Foundation Trust, and University Hospitals Leicester NHS Trust) collectively contributing 2 513 546 chest x-ray studies taken from a 13-year period (2006-19), which yielded 1 940 508 usable free-text radiological reports written by the contemporary assessing radiologist (collectively referred to as the "historic reporters") and 1 896 034 frontal images. Chest x-rays were labelled using a taxonomy of 37 findings by a custom-trained natural language processing (NLP) algorithm, X-Raydar-NLP, from the original free-text reports. X-Raydar-NLP was trained on 23 230 manually annotated reports and tested on 4551 reports from all hospitals. 1 694 921 labelled images from the training set and 89 238 from the validation set were then used to train a multi-label image classifier. Our algorithms were evaluated on three retrospective datasets: a set of exams sampled randomly from the full NHS dataset reported during clinical practice and annotated using NLP (n=103 328); a consensus set sampled from all six hospitals annotated by three expert radiologists (two independent annotators for each image and a third consultant to facilitate disagreement resolution) under research conditions (n=1427); and an independent dataset, MIMIC-CXR, consisting of NLP-annotated exams (n=252 374). Findings: X-Raydar achieved a mean AUC of 0·919 (SD 0·039) on the auto-labelled set, 0·864 (0·102) on the consensus set, and 0·842 (0·074) on the MIMIC-CXR test, demonstrating similar performance to the historic clinical radiologist reporters, as assessed on the consensus set, for multiple clinically important findings, including pneumothorax, parenchymal opacification, and parenchymal mass or nodules. On the consensus set, X-Raydar outperformed historical reporter balanced accuracy with significance on 27 of 37 findings, was non-inferior on nine, and inferior on one finding, resulting in an average improvement of 13·3% (SD 13·1) to 0·763 (0·110), including a mean 5·6% (13·2) improvement in critical findings to 0·826 (0·119). Interpretation: Our study shows that automated classification of chest x-rays under a comprehensive taxonomy can achieve performance levels similar to those of historical reporters and exhibit robust generalisation to external data. The open-sourced neural networks can serve as foundation models for further research and are freely available to the research community. Funding: Wellcome Trust.
-
Facial twitching: calcium or concussion conundrum? Hypocalcaemia in a young American football player masking an internal carotid artery dissectionA 30-year-old male American football player presented to the acute medical unit with left-hand and hemifacial spasms. History and examination revealed hemifacial spasms in keeping with seizure-like activity possibly due to symptomatic hypocalcaemia. Subsequent investigations revealed an adjusted calcium of 1.87 mmol/L and, hence, he was managed with intravenous calcium replacement. He presented two further times in a 1-month period, with subjective limb weakness, despite normal adjusted calcium. During his third admission, he developed slurred speech and a marked facial droop, with absence of power in the right upper limb. Imaging revealed acute and old infarctions in the left middle cerebral artery territory and appearances consistent with left internal carotid artery dissection. This presentation of arterial stroke is atypical but with potentially grave consequences if missed. There is limited literature on the presentation of hemifacial spasm, and its association with ischaemic or haemorrhagic stroke represents a key learning point. Keywords: calcium and bone; stroke.
-
Interventional radiology training in the UK: a view from within-a national survey.Objective: Interventional radiology (IR) training in the UK has evolved since recognition as a subspecialty in 2010 and introduction of a new curriculum in 2021. The changing landscape, increasing workload and COVID-19 have affected training. The purpose of this study was to review trainees' perspectives on training and develop strategies to further improve training. Methods: Online survey approved by the British Society of Interventional Radiology Council distributed to British Society of Interventional Radiology Trainee members between 9 March 22 and 25 March 2022. The survey was open to all UK based ST4-6 IR trainees and fellows. Descriptive and thematic analysis was undertaken. Results: 43 responses were received from 17/19 UK training regions. Females represented 10% (4/41) and 5% (2/43) less than full time (LTFT) trainees. 82% (31/38) felt their curriculum was suitable for their training and 28/38 (74%) were satisfied with IR training. Vascular IR, Interventional Oncology, paediatrics and stroke thrombectomy were identified as areas of training desiring improvement. 45% (18/40) stated exposure to IR led clinics and 17.5% (7/40) to IR led ward rounds. Only 6/38 (15.7%) received structured IR teaching at least once a month. Approximately, a third of respondents (13/38) stated training opportunities were significantly compromised secondary to COVID-19. Conclusion: This survey shows overall good satisfaction with IR training. However, improved training opportunities in vascular IR, interventional oncology, paediatric IR and stroke thrombectomy are required. In addition, access to clinics, ward rounds and protected time for research is needed to improve training quality.
-
Diagnostic pathways in Crohn's disease.The management of Crohn's disease (CD) is shifting from a stepwise, incremental approach based on symptom control to more aggressive early combined immunosuppression in an attempt to induce remission more rapidly and avoid long-term bowel damage. Accurately defining disease activity is a major challenge, as there is often a disconnect between symptomatology and underlying disease status. The role of imaging in CD has evolved such that it now plays a central role establishing the initial diagnosis, characterising disease phenotype, activity assessment, disease surveillance, and assessing response to therapy. Furthermore, the "treat-to-target" approach is being investigated in CD, with resolution of transmural inflammation on cross-sectional imaging being the treatment goal. In this review, we summarise the principal imaging techniques available to the radiologist, the key findings, and provide some guidance on the preferred imaging option in the diagnostic pathway. We consider the relative merits and drawbacks of each imaging technique before offering a brief discussion of some current developments and research avenues in CD imaging. We discuss how imaging may be useful in a "treat-to-target" approach. Finally, we highlight some practical considerations around service configuration and delivery to optimise imaging in CD in an accurate, cost-effective manner.
-
Imaging of Gastrointestinal Bleeding: An Update.no abstract available
-
Better safe than so ray: national survey of radiation protection amongst interventional radiology trainees in the United Kingdom.Objective: To establish the provision and use of radiation personal protective equipment (PPE) and dosimetry amongst UK interventional radiology (IR) trainees and highlight areas of improvement in order to enhance the radiation safety. Methods: A survey questionnaire was designed by members of the British Society of Interventional Radiology (BSIR) trainee committee via survey monkey and distributed to UK IR trainees via the BSIR membership mailing list, local representatives and Twitter. The survey was open from 04/01/2021 to 20/02/2021. Only IR trainees in years ST4 and above were included. Results: Of the 73 respondents, 62 qualified for analysis. Respondents (81% male) spent a median of 5.5 sessions (half day list) per week in the angiography suite and 58% (n=36) had difficulty finding appropriately sized lead aprons at least once a week. Overall 53% (n=33) had concerns about their radiation PPE. Furthermore 56% of trainees (n=35) experienced back pain among other symptoms attributed to wearing the lead aprons available to them. 77% (n=48) regularly wore lead glasses. For trainees requiring prescription glasses (n=22) overfit goggles were provided however 17 (77%) of these trainees felt the goggles compromised their ability to perform the procedure. Eye and finger dosimeters were used by 50% and 52% of respondents respectively. Compliance with body dosimetry was 99%. Conclusion: Provision of radiation PPE and dose monitoring for IR trainees is suboptimal, particularly access to adequate eye protection or suitably fitting leads. Based on the findings of this survey, recommendations have been made to promote the safety and radiation awareness of IR trainees.
-
Non-Hodgkin lymphoma causing hypopituitarism - can imaging help diagnosis and management?Non-Hodgkin lymphomas of the hypothalamus and pituitary are rare. They usually remain clinically silent until onset of compressive features affecting surrounding structures. When symptomatic, patients most commonly present with diabetes insipidus, headaches, ophthalmoplegia and/or bilateral hemianopia. We report a case of a 67-year-old Caucasian female with a history of B-cell lymphoma in complete remission. She presented with left oculomotor nerve palsy and was subsequently found to have a sellar/suprasellar mass lesion on MRI. Alongside hypocortisolism and hypogonadotropic hypogonadism, she developed transient diabetes insipidus during her illness. Her clinical course was characterized by rapid intracranial progression of the sellar mass. MR spectroscopy suggested a diagnosis of lymphoma. Diagnostic biopsy confirmed high-grade diffuse large B-cell CNS lymphoma; this changed the definitive management from surgical excision to chemotherapy. Despite treatment, she succumbed to her illness within 7 months of initial presentation. This case highlights the aggressive nature of CNS lymphomas and the need for a high index of suspicion in an unusual presentation of sellar/suprasellar mass lesions. Learning points: Novel imaging techniques such as MR spectroscopy might help to differentiate some brain tumours from pituitary macroadenomas, but these are not diagnostic.Tissue diagnosis with biopsy and histopathology is the gold standard for deciding management of pituitary fossa mass lesions with atypical presentation.