Automated pneumothorax triaging in chest X-rays in the New Zealand population using deep-learning algorithms.

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.

Curation of the CANDID-PTX Dataset with Free-Text Reports.

Supplemental material is available for this article. Keywords: Conventional Radiography, Thorax, Trauma, Ribs, Catheters, Segmentation, Diagnosis, Classification, Supervised Learning, Machine Learning © RSNA, 2021.

Exposure to harmful housing conditions is common in children admitted to Wellington Hospital.

AIM: To measure the prevalence of exposure to potentially modifiable risk factors in the homes of children hospitalised in Wellington. METHODS: Parents/caregivers of all children admitted to Wellington Public Hospital during a two-week period in July 2012 completed a standardised questionnaire in a face-to-face interview. The questionnaire collected sociodemographic, health and housing condition data. RESULTS: We interviewed parents/caregivers of 106 children, of whom 72% were aged 0-4 years. Respiratory conditions were the most common cause of admission. One third of parents noticed dampness and mould in their house, 50% stated that their house was colder than they preferred during the past month, 20% lived in uninsulated houses, 20% lived in overcrowded houses, and 38% were exposed to second hand smoke (SHS). Compared to New Zealand European (NZE) children, the odds ratios (OR) for Pacific children living in cold and overcrowded houses and being exposed to SHS were 14.0 (95%CI 3.0-66.0), 10.8 (95%CI 2.6-44.1) and 16.0 (95%CI 4.8-55.5) respectively. OR for Maori children living in cold and overcrowded houses and being exposed to SHS were 3.0 (95%CI 1.0-9.0), 6.8 (95%CI 1.6-30.1) and 8.0 (95%CI 2.5-28.6) respectively, compared to NZE children. The OR for children from deprived neighbourhoods (NZDep2006 areas 7-10) living in cold and overcrowded houses and being exposed to SHS were 4.1 (95%CI 1.8-9.6), 5.7 (95%CI 1.9-17.0) and 4.1 (95%CI 1.6-9.6) respectively. CONCLUSIONS: Among children admitted to Wellington Hospital there is a high prevalence of exposure to cold, damp and overcrowded houses and many children are exposed to SHS. Maori and Pacific children and children living in socioeconomically deprived areas are more likely than others to be exposed to these potential risk factors for childhood hospitalisation. This audit of child admissions could be repeated to provide surveillance of modifiable risk factors. A shortened version of the questionnaire could be used to screen children to identify those with harmful exposures in their home environment, provided suitable intervention programmes can be established.