Introduction: As a Quality Improvement initiative our department has held regular discrepancy meetings since 2003. We performed a retrospective analysis of the cases presented and identified the most common pattern of error. Methods: A total of 558 cases were referred for discussion over 92 months, and errors were classified as perceptual or interpretative. The most common patterns of error for each imaging modality were analysed, and the misses were scored by consensus as subtle or non-subtle. Results: Of 558 diagnostic errors, 447 (80%) were perceptual and 111 (20%) were interpretative errors. Plain radiography and computed tomography (CT) scans were the most frequent imaging modalities accounting for 246 (44%) and 241 (43%) of the total number of errors, respectively. In the plain radiography group 120 (49%) of the errors occurred in chest X-ray reports with perceptual miss of a lung nodule occurring in 40% of this subgroup. In the axial and appendicular skeleton missed fractures occurred most frequently, and metastatic bone disease was overlooked in 12 of 50 plain X-rays of the pelvis or spine. The majority of errors within the CT group were in reports of body scans with the commonest perceptual errors identified including 16 missed significant bone lesions, 14 cases of thromboembolic disease and 14 gastrointestinal tumours. Of the 558 errors, 312 (56%) were considered subtle and 246 (44%) non-subtle. Conclusion: Diagnostic errors are not uncommon and are most frequently perceptual in nature. Identification of the most common patterns of error has the potential to improve the quality of reporting by improving the search behaviour of radiologists.
Summary To determine the prevalence and spectrum of extrarenal findings in a screening population of potential living kidney donors undergoing renal Computed tomography angiography (CTA) and evaluate their impact on subsequent patient management and imaging costs. Two radiologists retrospectively reviewed 175 consecutive renal CTA's performed for assessment of potential living kidney donors. Extrarenal radiological findings were recorded and classified according to high, medium, or low importance based on clinical relevance and the need for further investigations and/or treatment. The cost of additional imaging examinations was calculated using 2002 Canadian (British Columbia) reimbursements. There were 73 extrarenal findings in 71/175 (40.6%) of the potential kidney donors in the study population. Findings were categorized as of high clinical importance in 18 (10.3%) cases, including lung lesions, bowel tumors, and liver tumors and as medium importance in 31 (17.7%). Twenty‐two (12.6%) individuals had findings categorized as low importance, probably of no clinical significance and requiring no follow‐up. Further potential evaluation of the 49 patients (28%) with highly and moderately significant extrarenal findings may require an additional $6137 (mean $35.1 per each case of all the screened patients). Transplantation of a kidney from a living donor is an excellent alternative to cadaveric allografts. Potential living kidney donors are a highly selected population of healthy individuals, screened for significant past or current medical conditions before undergoing CTA. Despite this screening, potentially significant extrarenal findings (classified as high or medium importance) were revealed in 28% of patients. These patients may require further investigations and/or treatment. The referring physician and patient should be aware of such potentially high probability, which may require further nontransplant related evaluation and treatment. This has medical, legal, economic, and ethical implications.
Introduction The primary aim was to develop convolutional neural network (CNN)‐based artificial intelligence (AI) models for pneumothorax classification and segmentation for automated chest X‐ray (CXR) triaging. A secondary aim was to perform interpretability analysis on the best‐performing candidate model to determine whether the model's predictions were susceptible to bias or confounding. Method A CANDID‐PTX dataset, that included 19,237 anonymized and manually labelled CXRs, was used for training and testing candidate models for pneumothorax classification and segmentation. Evaluation metrics for classification performance included Area under the receiver operating characteristic curve (AUC‐ROC), sensitivity and specificity, whilst segmentation performance was measured using mean Dice and true‐positive (TP)‐Dice coefficients. Interpretability analysis was performed using Grad‐CAM heatmaps. Finally, the best‐performing model was implemented for a triage simulation. Results The best‐performing model demonstrated a sensitivity of 0.93, specificity of 0.95 and AUC‐ROC of 0.94 in identifying the presence of pneumothorax. A TP‐Dice coefficient of 0.69 is given for segmentation performance. In triage simulation, mean reporting delay for pneumothorax‐containing CXRs is reduced from 9.8 ± 2 days to 1.0 ± 0.5 days (P‐value < 0.001 at 5% significance level), with sensitivity 0.95 and specificity of 0.95 given for the classification performance. Finally, interpretability analysis demonstrated models employed logic understandable to radiologists, with negligible bias or confounding in predictions. Conclusion AI models can automate pneumothorax detection with clinically acceptable accuracy, and potentially reduce reporting delays for urgent findings when implemented as triaging tools.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.