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.

Evidence for multiple bulbar and higher brain circuits processing sensory inputs from the respiratory system in humans.

KEY POINTS: Unpleasant respiratory sensations contribute to morbidity in pulmonary disease. In rodents, these sensations are processed by nodose and jugular vagal sensory neurons, two distinct cell populations that differentially project to the airways and brainstem. Whether similar differences exist in bronchopulmonary sensory pathways in humans is unknown. We use functional magnetic resonance imaging during inhalation of capsaicin and ATP, showing that airway nodose pathways project centrally to the nucleus of the solitary tract, whereas jugular pathways input into the trigeminal brainstem nuclei. We also show differences between the efficacy of nodose and jugular stimuli to evoke cough and activity in motor control regions of the brain. Our data suggest that humans have two distinct vagal sensory neural systems governing airway sensations and this may have implications for the development of new antitussive therapies. ABSTRACT: In rodents, nodose vagal sensory neurons preferentially innervate the distal airways and terminate centrally in the nucleus of the solitary tract. By contrast, jugular vagal sensory neurons preferentially innervate the proximal airways and terminate in the paratrigeminal nucleus in the dorsolateral medulla. This differential organization suggests distinct roles for nodose and jugular pathways in respiratory sensory processing. However, it is unknown whether bronchopulmonary afferent pathways are similarly arranged in humans. We set out to investigate this using high resolution brainstem and whole brain functional magnetic resonance imaging in healthy human participants when they were inhaling stimuli known to differentially activate nodose and jugular pathways. Inhalation of capsaicin or ATP evoked respiratory sensations described as an urge-to-cough, although ATP was significantly less effective compared to capsaicin at evoking the motor act of coughing. The nodose and jugular neuron stimulant capsaicin increased blood oxygen level-dependent (BOLD) signals extending across the dorsomedial and dorsolateral medulla, encompassing regions containing both the nucleus of the solitary tract and the paratrigeminal nucleus. By contrast, at perceptually comparable stimulus intensities, the nodose-selective stimulant ATP resulted in BOLD signal intensity changes that were confined to the area of the nucleus of the solitary tract. During whole brain imaging, capsaicin demonstrated a wider distributed network of activity compared to ATP, with significantly increased activity in regions involved with motor control functions. These data suggest that functional and neuroanatomical differences in bronchopulmonary nodose and jugular sensory pathway organization are conserved in humans and also that this has implications for understanding the neurobiological mechanisms underpinning cough.

Postnatal germ cell development during first 18 months of life in testes from boys with non-syndromic cryptorchidism and complete or partial androgen insensitivity syndrome.

INTRODUCTION: Neonatal testicular germ cells/gonocytes, transform into stem cells for spermatogenesis during 'minipuberty', driving change in timing of surgery. This study examined gonocyte transformation in cryptorchid testes in children ≤18 months of age with unilateral, bilateral undescended testes (UDT), complete or partial androgen insensitivity syndrome (CAIS, PAIS) [3,4]. MATERIAL AND METHODS: Testicular biopsies were taken from patients with unilateral or bilateral UDT, PAIS or CAIS, aged 10 days-18 months. These testicular sections underwent immunohistochemistry with antibodies (Oct4, Ki67, C-Kit, Sox9) followed by confocal imaging, cell counting and statistical analysis. RESULTS: Both Sertoli cells/tubule and germ cells (GC)/tubule decreased with age, and % empty tubules (no GC) increased with age but with no significant differences between patient groups. Oct4+ germ cells/tubule decreased with age. There are some GCs and Sertoli cells proliferating during the first year and most proliferating Oct4+ germ cells (Oct4+/Ki67+) were located off tubular basement membrane. CONCLUSION: Our study showed that Oct4 expression gradually decreased after minipuberty and transformation into spermatogonia. Germ cells and Sertoli cells undergo mitosis during the first 12 months although not abundantly. We propose that Oct4+ gonocyte transformation into spermatogonia via proliferation and migration to the basement membrane may be delayed in UDT.