IntroductionThe gold standard in general radiography is to place a radiopaque anatomical side marker in the field of view for each radiographic image prior to exposure. The advent of digital radiography has allowed for anatomical side markers to be digitally added to films as part of post‐processing. The aim of this audit was to identify whether general X‐ray images performed in a tertiary Women's and Children's Hospital were being appropriately annotated with a definitive side marker, and to identify factors that may contribute to inappropriately labelled images.MethodsFour hundred images from 201 patients’ examinations occurring within a randomly selected time period were assessed to ascertain whether radiographic anatomical side markers were visible when images were viewed via the hospitals main viewing platform. The audit occurred in January 2014. The scope included both mobile and in‐department general X‐ray examinations, with the patient age range extending from 1 day to 18 years.ResultsOf the 400 images evaluated, 88 (22%) were found to have a lead marker that matched the anatomy being imaged within the primary beam; 289 (72.3%) images contained a correct digital marker inserted as part of the post‐processing of the image. In total, 377 (94.2%) images were appropriately marked. Of the 23 (5.8%) images not marked correctly, 22 images had no marker and 1 was incorrectly marked with a digital marker. There was a noticeable relationship between absent anatomical markers and chest X‐rays performed outside of the medical imaging department.ConclusionsWhile it is encouraging that the majority of the images assessed were correctly annotated, with only a small number of missing markers, there are opportunities for further improvement. The audit findings suggest that reduced access to lead markers influences marker use. Strategies that may improve compliance at an individual level include distribution of personalised anatomical side markers, and targeted staff education sessions. At a department level, regular audits and monitoring should be encouraged.

No abstract

The detection and removal of poor-quality data in a training set is crucial to achieve high-performing AI models. In healthcare, data can be inherently poor-quality due to uncertainty or subjectivity, but as is often the case, the requirement for data privacy restricts AI practitioners from accessing raw training data, meaning manual visual verification of private patient data is not possible. Here we describe a novel method for automated identification of poor-quality data, called Untrainable Data Cleansing. This method is shown to have numerous benefits including protection of private patient data; improvement in AI generalizability; reduction in time, cost, and data needed for training; all while offering a truer reporting of AI performance itself. Additionally, results show that Untrainable Data Cleansing could be useful as a triage tool to identify difficult clinical cases that may warrant in-depth evaluation or additional testing to support a diagnosis.

Introduction: Diagnostic Reference Levels (DRL) of procedures involving ionizing radiation are important tools to optimizing radiation doses delivered to patients and in identifying cases where the levels of doses are unusually high. This is particularly important for paediatric patients undergoing computed tomography (CT) examinations as these examinations are associated with relatively high-dose. Methods: Paediatric CT studies, performed at our institution from January 2010 to March 2014, have been retrospectively analysed to determine the 75th and 95th percentiles of both the volume computed tomography dose index (CTDI vol ) and dose-length product (DLP) for the most commonly performed studies to: 1 establish local diagnostic reference levels for paediatric computed tomography examinations performed at our institution, 2 benchmark our DRL with national and international published paediatric values, and 3 determine the compliance of CT radiographer with established protocols. Results: The derived local 75th percentile DRL have been found to be acceptable when compared with those published by the Australian National Radiation Dose Register and two national children's hospitals, and at the international level with the National Reference Doses for the UK. The 95th percentiles of CTDIvol for the various CT examinations have been found to be acceptable values for the CT scanner Dose-Check Notification. Benchmarking CT radiographers shows that they follow the set protocols for the various examinations without significant variations in the machine setting factors. Conclusion: The derivation of DRL has given us the tool to evaluate and improve the performance of our CT service by improved compliance and a reduction in radiation dose to our paediatric patients. We have also been able to benchmark our performance with similar national and international institutions.

Objective: To determine the influence of antenatal ultrasound on the management of exomphalos. Methods: Retrospective case note review of 23 fetuses and infants referred to our institution with either a pre- or postnatal diagnosis of exomphalos over a 7-year period. Results: There were 21 cases of exomphalos of which 18 were correctly diagnosed on antenatal ultrasound by 18 weeks’ gestation. There were 2 false-positives and 3 false-negatives, including 1 case of amniotic band syndrome with an abdominal wall defect and 1 morphologically normal fetus. Associated anomalies were correctly identified in 12 but incorrectly reported in 8. Maternal serum α-fetoprotein levels were abnormal in 61% of cases of abdominal wall defects in this series. Amniocentesis was performed in 12 and cordocentesis in 1. There were 13 terminations, including 2 trisomy 18s and 1 trisomy 13. Two fetal deaths followed amniocentesis. Of the 10 live births, 9 had their exomphalos repaired with a 1-year survival rate of 89%. Prenatal diagnosis did not appear to influence outcome. Conclusions: Antenatal ultrasound diagnosed 86% of cases of exomphalos and correctly reported 67% of associated anomalies. Amniocentesis may have led to the death of 1 morphologically normal fetus.