Post-mortem skeletal survey (PMSS) versus post-mortem computed tomography (PMCT) for the detection of corner metaphyseal lesions (CML) in children.

OBJECTIVES: Corner metaphyseal lesions (CMLs) are specific for child abuse but challenging to detect on radiographs. The accuracy of CT for CML detection is unknown. Our aim was to compare diagnostic accuracy for CML detection on post-mortem skeletal surveys (PMSS, plain radiography) versus post-mortem CT (PMCT). METHODS: A 10-year retrospective review was performed at a children's hospital for patients having PMSS, PMCT and histopathological correlation (reference standard) for suspected CMLs. Twenty-four radiologists independently reported the presence or absence of CMLs in all cases in a blinded randomised cross-over design across two rounds. Logistic regression models were used to compare accuracy between modalities. RESULTS: Twenty CMLs were reviewed for each of the 10 subjects (200 metaphyses in all). Among them, 20 CMLs were confirmed by bone histopathology. Sensitivity for these CMLs was significantly higher for PMSS (69.6%, 95% CI 61.7 to 76.7) than PMCT (60.5%, 95% CI 51.9 to 68.6). Using PMSS for detection of CMLs would yield one extra correct diagnosis for every 11.1 (95% CI 6.6 to 37.0) fractured bones. In contrast, specificity was higher on PMCT (92.7%, 95% CI 90.3 to 94.5) than PMSS (90.5%, 95% CI 87.6 to 92.8) with an absolute difference of 2.2% (95% CI 1.0 to 3.4, p < 0.001). More fractures were reported collectively by readers on PMSS (785) than on PMCT (640). CONCLUSION: PMSS remains preferable to PMCT for CML evaluation. Any investigation of suspected abuse or unexplained deaths should include radiographs of the limbs to exclude CMLs. CLINICAL RELEVANCE STATEMENT: In order to avoid missing evidence that could indicate child abuse as a contributory cause for death in children, radiographs of the limbs should be performed to exclude CMLs, even if a PMCT is being acquired. KEY POINTS: • Corner metaphyseal lesions (CMLs) are indicative for abuse, but challenging to detect. Skeletal surveys (i.e. radiographs) are standard practice; however, accuracy of CT is unknown. • Sensitivity for CML detection on radiographs is significantly higher than CT. • Investigation of unexplained paediatric deaths should include radiographs to exclude CMLs even if CT is also being performed.

"How I would like AI used for my imaging": children and young persons' perspectives.

OBJECTIVES: Artificial intelligence (AI) tools are becoming more available in modern healthcare, particularly in radiology, although less attention has been paid to applications for children and young people. In the development of these, it is critical their views are heard. MATERIALS AND METHODS: A national, online survey was publicised to UK schools, universities and charity partners encouraging any child or young adult to participate. The survey was "live" for one year (June 2022 to 2023). Questions about views of AI in general, and in specific circumstances (e.g. bone fractures) were asked. RESULTS: One hundred and seventy-one eligible responses were received, with a mean age of 19 years (6-23 years) with representation across all 4 UK nations. Most respondents agreed or strongly agreed they wanted to know the accuracy of an AI tool that was being used (122/171, 71.3%), that accuracy was more important than speed (113/171, 66.1%), and that AI should be used with human oversight (110/171, 64.3%). Many respondents (73/171, 42.7%) felt AI would be more accurate at finding problems on bone X-rays than humans, with almost all respondents who had sustained a missed fracture strongly agreeing with that sentiment (12/14, 85.7%). CONCLUSIONS: Children and young people in our survey had positive views regarding AI, and felt it should be integrated into modern healthcare, but expressed a preference for a "medical professional in the loop" and accuracy of findings over speed. Key themes regarding information on AI performance and governance were raised and should be considered prior to future AI implementation for paediatric healthcare. CLINICAL RELEVANCE STATEMENT: Artificial intelligence (AI) integration into clinical practice must consider all stakeholders, especially paediatric patients who have largely been ignored. Children and young people favour AI involvement with human oversight, seek assurances for safety, accuracy, and clear accountability in case of failures. KEY POINTS: Paediatric patient's needs and voices are often overlooked in AI tool design and deployment. Children and young people approved of AI, if paired with human oversight and reliability. Children and young people are stakeholders for developing and deploying AI tools in paediatrics.

Pre- and post-COVID-19 gender trends in authorship for paediatric radiology articles worldwide: a systematic review.

BACKGROUND: Gender inequalities in academic medicine persist despite progress over the past decade. Evidence-based targeted interventions are needed to reduce gender inequalities. OBJECTIVE: This systematic review aimed to investigate the impact of COVID-19 on gender trends in authorship of paediatric radiology research worldwide. MATERIALS AND METHODS: This prospectively registered, PRISMA-compliant systematic review searched the following databases: PubMed, MEDLINE, Web of Science, and Scopus from January 1, 2018, to May 29, 2023, with no restrictions on country of origin. Screening and data extraction occurred independently and in duplicate. Gender of first, last, and corresponding authors were determined using an artificial intelligence-powered, validated, multinational database ( www.genderize.io ). Two time periods were categorised according to the Johns Hopkins Center for Systems Science and Engineering: pre-COVID (prior to March 2020) and peak and post-COVID (March 2020 onwards). One-sample binomial testing was used to analyse proportion of authorship based on gender. Categorical variables were described as frequencies and percentages, and compared using testing chi-square or Fisher exact testing, with a threshold of P<0.05 representing statistical significance. RESULTS: In total, 922 articles were included with 39 countries represented. A statistically significant difference in authorship based on gender persisted during the peak and post-COVID time period (March 2020 onwards) where women represented a statistically significant lower proportion of last (35.5%) and corresponding (42.7%) authors (P<0.001, P=0.001, respectively). Statistically significant differences for first authors were not found in either period (P=0.08 and P=0.48). CONCLUSION: This study identifies differences in gender trends for authorship in paediatric radiology research worldwide. Future efforts to increase authorship by women are needed.

Artificial Intelligence in Paediatric Tuberculosis.

Tuberculosis (TB) continues to be a leading cause of death in children despite global efforts focused on early diagnosis and interventions to limit the spread of the disease. This challenge has been made more complex in the context of the coronavirus pandemic, which has disrupted the "End TB Strategy" and framework set out by the World Health Organization (WHO). Since the inception of artificial intelligence (AI) more than 60 years ago, the interest in AI has risen and more recently we have seen the emergence of multiple real-world applications, many of which relate to medical imaging. Nonetheless, real-world AI applications and clinical studies are limited in the niche area of paediatric imaging. This review article will focus on how AI, or more specifically deep learning, can be applied to TB diagnosis and management in children. We describe how deep learning can be utilised in chest imaging to provide computer-assisted diagnosis to augment workflow and screening efforts. We also review examples of recent AI applications for TB screening in resource constrained environments and we explore some of the challenges and the future directions of AI in paediatric TB.

Micro-CT yields high image quality in human fetal post-mortem imaging despite maceration.

BACKGROUND: Current clinical post-mortem imaging techniques do not provide sufficiently high-resolution imaging for smaller fetuses after pregnancy loss. Post-mortem micro-CT is a non-invasive technique that can deliver high diagnostic accuracy for these smaller fetuses. The purpose of the study is to identify the main predictors of image quality for human fetal post-mortem micro-CT imaging. METHODS: Human fetuses were imaged using micro-CT following potassium tri-iodide tissue preparation, and axial head and chest views were assessed for image quality on a Likert scale by two blinded radiologists. Simple and multivariable linear regression models were performed with demographic details, iodination, tissue maceration score and imaging parameters as predictor variables. RESULTS: 258 fetuses were assessed, with median weight 41.7 g (2.6-350 g) and mean gestational age 16 weeks (11-24 weeks). A high image quality score (> 6.5) was achieved in 95% of micro-CT studies, higher for the head (median = 9) than chest (median = 8.5) imaging. The strongest negative predictors of image quality were increasing maceration and body weight (p < 0.001), with number of projections being the best positive imaging predictor. CONCLUSIONS: High micro-CT image quality score is achievable following early pregnancy loss despite fetal maceration, particularly in smaller fetuses where conventional autopsy may be particularly challenging. These findings will help establish clinical micro-CT imaging services, addressing the need for less invasive fetal autopsy methods.

Imaging findings of multisystem inflammatory syndrome in children associated with COVID-19.

BACKGROUND: A hyperinflammatory immune-mediated shock syndrome has been recognised in children exposed to the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that causes coronavirus disease 2019 (COVID-19). OBJECTIVE: To describe typical imaging findings in children with multisystem inflammatory syndrome associated with COVID-19. MATERIALS AND METHODS: During the first wave of the COVID-19 pandemic, imaging studies and clinical data from children treated for multisystem inflammatory syndrome were collected from multiple centres. Standardised case templates including demographic, biochemical and imaging information were completed by participating centres and reviewed by paediatric radiologists and paediatricians. RESULTS: We included 37 children (21 boys; median age 8.0 years). Polymerase chain reaction (PCR) testing was positive for SARS-CoV-2 in 15/37 (41%) children and immunoglobulins in 13/19 children (68%). Common clinical presentations were fever (100%), abdominal pain (68%), rash (54%), conjunctivitis (38%) and cough (32%). Thirty-three children (89%) showed laboratory or imaging findings of cardiac involvement. Thirty of the 37 children (81%) required admission to the intensive care unit, with good recovery in all cases. Chest radiographs demonstrated cardiomegaly in 54% and signs of pulmonary venous hypertension/congestion in 73%. The most common chest CT abnormalities were ground-glass and interstitial opacities (83%), airspace consolidation (58%), pleural effusion (58%) and bronchial wall thickening (42%). Echocardiography revealed impaired cardiac function in half of cases (51%) and coronary artery abnormalities in 14%. Cardiac MRI showed myocardial oedema in 58%, pericardial effusion in 42% and decreased left ventricular function in 25%. Twenty children required imaging for abdominal symptoms, the commonest abnormalities being free fluid (71%) and terminal ileum wall thickening (57%). Twelve children underwent brain imaging, showing abnormalities in two cases. CONCLUSION: Children with multisystem inflammatory syndrome showed pulmonary, cardiac, abdominal and brain imaging findings, reflecting the multisystem inflammatory disease. Awareness of the imaging features of this disease is important for early diagnosis and treatment.

Maceration determines diagnostic yield of fetal and neonatal whole body post-mortem ultrasound.

OBJECTIVES: To determine factors in nondiagnostic fetal and neonatal post-mortem ultrasound (PMUS) examinations. METHODS: All fetal and neonatal PMUS examinations were included over a 5-year study period (2014-2019). Nondiagnostic image quality by body parts (brain, spine, thorax, cardiac, and abdomen) was recorded and correlated with patient variables. Descriptive statistics and logistic regression analyses were performed to identify significant factors for nondiagnostic studies. RESULTS: Two hundred sixty-five PMUS examinations were included, with median gestational age of 22 weeks (12-42 wk), post-mortem weight of 363 g (16-4033 g), and post-mortem interval of 8 days (0-39 d). Diagnostic imaging quality was achieved for 178/265 (67.2%) studies. It was high for abdominal (263/265, 99.2%), thoracic (264/265, 99.6%), and spine (265/265, 100%) but lower for brain (210/265, 79.2%) and cardiac imaging (213/265, 80.4%). Maceration was the best overall predictor for nondiagnostic imaging quality (P < .0001). Post-mortem fetal weight was positively associated with cardiac (P = .0133) and negatively associated with brain imaging quality (P = .0002). Post-mortem interval was not a significant predictor. CONCLUSIONS: Fetal maceration was the best predictor for nondiagnostic PMUS, particularly for brain and heart. Fetuses with marked maceration and suspected cardiac or brain anomalies should be prioritised for post-mortem MRI.

Thoracic imaging of coronavirus disease 2019 (COVID-19) in children: a series of 91 cases.

BACKGROUND: Pulmonary infection with SARS-CoV-2 virus (severe acute respiratory syndrome coronavirus 2; COVID-19) has rapidly spread worldwide to become a global pandemic. OBJECTIVE: To collect paediatric COVID-19 cases worldwide and to summarize both clinical and imaging findings in children who tested positive on polymerase chain reaction testing for SARS-CoV-2. MATERIALS AND METHODS: Data were collected by completion of a standardised case report form submitted to the office of the European Society of Paediatric Radiology from March 12 to April 8, 2020. Chest imaging findings in children younger than 18 years old who tested positive on polymerase chain reaction testing for SARS-CoV-2 were included. Representative imaging studies were evaluated by multiple senior paediatric radiologists from this group with expertise in paediatric chest imaging. RESULTS: Ninety-one children were included (49 males; median age: 6.1 years, interquartile range: 1.0 to 13.0 years, range: 9 days-17 years). Most had mild symptoms, mostly fever and cough, and one-third had coexisting medical conditions. Eleven percent of children presented with severe symptoms and required intensive unit care. Chest radiographs were available in 89% of patients and 10% of them were normal. Abnormal chest radiographs showed mainly perihilar bronchial wall thickening (58%) and/or airspace consolidation (35%). Computed tomography (CT) scans were available in 26% of cases, with the most common abnormality being ground glass opacities (88%) and/or airspace consolidation (58%). Tree in bud opacities were seen in 6 of 24 CTs (25%). Lung ultrasound and chest magnetic resonance imaging were rarely utilized. CONCLUSION: It seems unnecessary to perform chest imaging in children to diagnose COVID-19. Chest radiography can be used in symptomatic children to assess airway infection or pneumonia. CT should be reserved for when there is clinical concern to assess for possible complications, especially in children with coexisting medical conditions.

Postmortem microfocus computed tomography for early gestation fetuses: a validation study against conventional autopsy.

BACKGROUND: Perinatal autopsy provides useful clinical information in up to 40% of cases. However, there is a substantial unmet clinical need with regards to postmortem investigation of early gestation fetal loss for parents for whom standard autopsy is either not available or not acceptable. Parents dislike the invasive nature of autopsy, but current clinical imaging techniques do not provide high-enough imaging resolution in small fetuses. We hypothesized that microfocus computed tomography, which is a rapid high-resolution imaging technique, could give accurate diagnostic imaging after early gestation fetal loss. OBJECTIVE: The objective of the study was to evaluate the diagnostic accuracy of microfocus computed tomography for noninvasive human fetal autopsy for early gestation fetuses, with the use of conventional autopsy as the reference standard. STUDY DESIGN: We compared iodinated whole body microfocus computed tomography in 20 prospectively recruited fetuses (11-21 weeks gestation from 2 centers) with conventional autopsy in a double-blinded manner for a main diagnosis and findings in specific body organs. Fetuses were prepared with 10% formalin/potassium tri-iodide. Images were acquired with a microfocus computed tomography scanner with size-appropriate parameters. Images were evaluated independently by 2 pediatric radiologists, who were blinded to formal perinatal autopsy results, across 40 individual indices to reach consensus. The primary outcome was agreement between microfocus computed tomography and conventional autopsy for overall diagnosis. RESULTS: Postmortem whole body fetal microfocus computed tomography gave noninvasive autopsy in minutes, at a mean resolution of 27µm, with high diagnostic accuracy in fetuses at <22 weeks gestation. Autopsy demonstrated that 13 of 20 fetuses had structural abnormalities, 12 of which were also identified by microfocus computed tomography (92.3%). Overall, microfocus computed tomography agreed with overall autopsy findings in 35 of 38 diagnoses (15 true positive, 18 true negative; sensitivity 93.8% [95% confidence interval, 71.7-98.9%], specificity 100% [95% confidence interval, 82.4-100%]), with 100% agreement for body imaging diagnoses. Furthermore, after removal of nondiagnostic indices, there was agreement for 700 of 718 individual body organ indices that were assessed on microfocus computed tomography and autopsy (agreement, 97.5%; 95% confidence interval, 96.1-98.4%), with no overall differences between fetuses at ≤14 or >14 weeks gestation (agreement, 97.2% and 97.9%, respectively). Within first-trimester fetal loss cases (<14 weeks gestation), microfocus computed tomography analysis yielded significantly fewer nondiagnostic indices than autopsy examination (22/440 vs 48/348, respectively; P<.001). CONCLUSION: Postmortem whole-body fetal microfocus computed tomography gives noninvasive, detailed anatomic examinations that are achieved in minutes at high resolution. Microfocus computed tomography may be preferable to magnetic resonance imaging in early gestation fetuses and may offer an acceptable method of examination after fetal loss for parents who decline invasive autopsy. This will facilitate autopsy and subsequent discussions between medical professionals who are involved in patient care and counselling for future pregnancies.

Imaging of the hip in juvenile idiopathic arthritis.

Hip involvement is common and estimated to occur in approximately 35-63% of children with juvenile idiopathic arthritis (JIA). It is more prevalent in the aggressive systemic subtypes, with irreversible changes occurring as early as within 5 years of diagnosis. Whilst clinical parameters and joint examination can be useful for assessing disease severity, subclinical disease is known to exist and delayed treatment may herald a lifetime of disability and pain. Early recognition of JIA changes is therefore crucial in determining treatment options. Validated scoring systems in the radiologic assessment of the hip for clinical drug trials may inform treatment outcomes, although robust tools for analysis are still lacking. This review article details the modalities utilised for imaging the hip in children with JIA with particular efforts focused upon reliability and validity in their assessment of joint disease. We conclude with a short literature review on the potential future techniques being developed for hip joint imaging in JIA.

Achondroplasia: Really rhizomelic?

Achondroplasia is the most common form of short limb dwarfism in humans. The shortening of the limb lengths in achondroplasia is widely described as "rhizomelic." While this appearance may be convincing clinically, the description is not necessarily true or helpful radiologically. The aims of this study, were therefore, to determine whether rhizomelic shortening is a true feature of achondroplasia at diagnosis in infancy. Humeral, radial, femoral, and tibial diaphyseal lengths were recorded by two independent observers from 22 skeletal surveys of infants with achondroplasia and compared with 150 normal age-matched control subjects. Upper and lower limb bone length ratios (radial/humeral and tibial/femoral lengths, respectively) in both groups were compared using an unpaired t-test. Mean upper limb length ratios were statistically higher within the achondroplasia group at 0.87 ± 0.04 (n = 22, mean age 70 ± 94 days) compared to normal controls at 0.79 ± 0.02 (n = 150, mean age 113 days ± 88 days; P < 0.0001). Lower limb length ratios were not significantly different between groups (0.84 ± 0.04 vs. 0.83 ± 0.02, P = 0.46). There was good inter-observer agreement of limb length measurements, with an average measurement difference of 0.1 ± 1.4 mm. In conclusion, infants with achondroplasia demonstrate statistically significant rhizomelic shortening within the upper limbs, but not lower limbs at diagnosis, compared to normal controls. The term "rhizomelic shortening" in relation to achondroplasia should be reserved when describing upper limb proportions. © 2016 Wiley Periodicals, Inc.

Can artificial intelligence pass the Fellowship of the Royal College of Radiologists examination? Multi-reader diagnostic accuracy study.

OBJECTIVE: To determine whether an artificial intelligence candidate could pass the rapid (radiographic) reporting component of the Fellowship of the Royal College of Radiologists (FRCR) examination. DESIGN: Prospective multi-reader diagnostic accuracy study. SETTING: United Kingdom. PARTICIPANTS: One artificial intelligence candidate (Smarturgences, Milvue) and 26 radiologists who had passed the FRCR examination in the preceding 12 months. MAIN OUTCOME MEASURES: Accuracy and pass rate of the artificial intelligence compared with radiologists across 10 mock FRCR rapid reporting examinations (each examination containing 30 radiographs, requiring 90% accuracy rate to pass). RESULTS: When non-interpretable images were excluded from the analysis, the artificial intelligence candidate achieved an average overall accuracy of 79.5% (95% confidence interval 74.1% to 84.3%) and passed two of 10 mock FRCR examinations. The average radiologist achieved an average accuracy of 84.8% (76.1-91.9%) and passed four of 10 mock examinations. The sensitivity for the artificial intelligence was 83.6% (95% confidence interval 76.2% to 89.4%) and the specificity was 75.2% (66.7% to 82.5%), compared with summary estimates across all radiologists of 84.1% (81.0% to 87.0%) and 87.3% (85.0% to 89.3%). Across 148/300 radiographs that were correctly interpreted by >90% of radiologists, the artificial intelligence candidate was incorrect in 14/148 (9%). In 20/300 radiographs that most (>50%) radiologists interpreted incorrectly, the artificial intelligence candidate was correct in 10/20 (50%). Most imaging pitfalls related to interpretation of musculoskeletal rather than chest radiographs. CONCLUSIONS: When special dispensation for the artificial intelligence candidate was provided (that is, exclusion of non-interpretable images), the artificial intelligence candidate was able to pass two of 10 mock examinations. Potential exists for the artificial intelligence candidate to improve its radiographic interpretation skills by focusing on musculoskeletal cases and learning to interpret radiographs of the axial skeleton and abdomen that are currently considered "non-interpretable."

Review of study reporting guidelines for clinical studies using artificial intelligence in healthcare.

High-quality research is essential in guiding evidence-based care, and should be reported in a way that is reproducible, transparent and where appropriate, provide sufficient detail for inclusion in future meta-analyses. Reporting guidelines for various study designs have been widely used for clinical (and preclinical) studies, consisting of checklists with a minimum set of points for inclusion. With the recent rise in volume of research using artificial intelligence (AI), additional factors need to be evaluated, which do not neatly conform to traditional reporting guidelines (eg, details relating to technical algorithm development). In this review, reporting guidelines are highlighted to promote awareness of essential content required for studies evaluating AI interventions in healthcare. These include published and in progress extensions to well-known reporting guidelines such as Standard Protocol Items: Recommendations for Interventional Trials-AI (study protocols), Consolidated Standards of Reporting Trials-AI (randomised controlled trials), Standards for Reporting of Diagnostic Accuracy Studies-AI (diagnostic accuracy studies) and Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis-AI (prediction model studies). Additionally there are a number of guidelines that consider AI for health interventions more generally (eg, Checklist for Artificial Intelligence in Medical Imaging (CLAIM), minimum information (MI)-CLAIM, MI for Medical AI Reporting) or address a specific element such as the 'learning curve' (Developmental and Exploratory Clinical Investigation of Decision-AI) . Economic evaluation of AI health interventions is not currently addressed, and may benefit from extension to an existing guideline. In the face of a rapid influx of studies of AI health interventions, reporting guidelines help ensure that investigators and those appraising studies consider both the well-recognised elements of good study design and reporting, while also adequately addressing new challenges posed by AI-specific elements.

Transarterial catheter embolisation for an unusual cause of upper gastrointestinal haemorrhage.

A 72-year-old man with a history of gallstones, and complex cardiac and endocrinological comorbidities, presented with severe abdominal pain and melaena. CT mesenteric angiogram showed a cystic artery pseudoaneurysm and gallbladder distended by haematoma. Subsequent mesenteric angiography confirmed a cystic artery pseudoaneurysm, which was successfully embolised with microcoils. The patient made a rapid recovery and was discharged after 3 days.

Delays and errors in abnormal chest radiograph follow-up: a systems approach to promoting patient safety in radiology.

RATIONALE, AIMS AND OBJECTIVES: This study aimed to apply the 'systems approach' to patient safety in order to identify causes for delays and errors in lung cancer diagnoses following an abnormal chest radiograph. METHODS: In the first part of this study, the systems approach to patient safety was comprehensively reviewed by three radiologists and seven patient safety experts. In the second part of this study, a retrospective review was performed of all patients referred to the lung cancer multidisciplinary team (MDT) meeting over a 1-year period. All abnormal chest radiograph reports were examined and a root-cause analysis performed of cases where errors and delays in diagnoses were deemed to have occurred. RESULTS: A total of 124 cases were reviewed, of which 36 (29%) patients had an abnormal preceding chest radiograph prior to MDT referral. In six cases, serious errors from delay and lack of follow-up were identified. These are analysed and discussed in detail in this article. Application of the systems approach to each case identified poor communication and lack of clinical action as prime causes. CONCLUSIONS: Both reporting radiologists and referring clinicians have a responsibility to ensure appropriate action following an abnormal chest radiograph. The main error lies in communication between the referring clinicians and the radiologists. Direct electronic communication is potentially a more robust method to overcome this.