Deep learning for classification of pediatric chest radiographs by WHO's standardized methodology.

Chen Y ，Roberts CS ，Ou W ，Petigara T ，Goldmacher GV ，Fancourt N ，Knoll MD ... -  《PLoS One》

A comprehensive segmentation of chest X-ray improves deep learning-based WHO radiologically confirmed pneumonia diagnosis in children.

To investigate a comprehensive segmentation of chest X-ray (CXR) in promoting deep learning-based World Health Organization's (WHO) radiologically confirmed pneumonia diagnosis in children. A total of 4400 participants between January 2016 and June 2021were identified for a cross-sectional study and divided into primary endpoint pneumonia (PEP), other infiltrates, and normal groups according to WHO's diagnostic criteria. The CXR was divided into six segments of left lung, right lung, mediastinum, diaphragm, ext-left lung, and ext-right lung by adopting the RA-UNet. To demonstrate the benefits of lung field segmentation in pneumonia diagnosis, the segmented images and images that were not segmented, which constituted seven segmentation combinations, were fed into the CBAM-ResNet under a three-category classification comparison. The interpretability of the CBAM-ResNet for pneumonia diagnosis was also performed by adopting a Grad-CAM module. The RA-UNet achieved a high spatial overlap between manual and automatic segmentation (averaged DSC = 0.9639). The CBAM-ResNet when fed with the six segments achieved superior three-category diagnosis performance (accuracy = 0.8243) over other segmentation combinations and deep learning models under comparison, which was increased by around 6% in accuracy, precision, specificity, sensitivity, F1-score, and around 3% in AUC. The Grad-CAM could capture the pneumonia lesions more accurately, generating a more interpretable visualization and enhancing the superiority and reliability of our study in assisting pediatric pneumonia diagnosis. The comprehensive segmentation of CXR could improve deep learning-based pneumonia diagnosis in childhood with a more reasonable WHO's radiological standardized pneumonia classification instead of conventional dichotomous bacterial pneumonia and viral pneumonia. The comprehensive segmentation of chest X-ray improves deep learning-based WHO confirmed pneumonia diagnosis in children, laying a strong foundation for the potential inclusion of computer-aided pediatric CXR readings in precise classification of pneumonia and PCV vaccine trials efficacy in children. • The chest X-ray was comprehensively segmented into six anatomical structures of left lung, right lung, mediastinum, diaphragm, ext-left lung, and ext-right lung. • The comprehensive segmentation improved the three-category classification of primary endpoint pneumonia, other infiltrates, and normal with an increase by around 6% in accuracy, precision, specificity, sensitivity, F1-score, and around 3% in AUC. • The comprehensive segmentation gave rise to a more accurate and interpretable visualization results in capturing the pneumonia lesions.

Li Y ，Zhang L ，Yu H ，Wang J ，Wang S ，Liu J ，Zheng Q ... -  《-》

Using deep-learning techniques for pulmonary-thoracic segmentations and improvement of pneumonia diagnosis in pediatric chest radiographs.

To evaluate the efficacy of a deep-learning model to segment the lung and thorax regions in pediatric chest X-rays (CXRs). Validating the diagnosis of bacterial or viral pneumonia could be improved after lung segmentation. A clinical-pediatric CXR set including 1351 patients was proposed to develop a deep-learning model for the pulmonary-thoracic segmentations. Model performance was evaluated by Jaccard's similarity coefficient (JSC) and Dice's coefficient (DC). Two adult CXR sets were used to assess the model's generalizability. According to the pulmonary-thoracic ratio, Pearson's correlation coefficient and the Bland-Altman plot were generated to demonstrate the correlation and agreement between manual and automatic segmentations. The receiver operating characteristic curves and areas under the curve (AUCs) were used to compare the pneumonia classification performance based on the lung-extracted images with that based on the original images. The model achieved JSCs of 0.910 and 0.950, DCs of 0.948 and 0.974 for lung and thorax segmentations, respectively. Pearson's r = 0.96, P < .0001. In the Bland-Altman plot, the mean difference was 0.0025 with a 95% confidence interval of (-0.0451, 0.0501). For testing with two adult CXR sets, the JSCs were 0.903 and 0.888, respectively, while the DCs were 0.948 and 0.937, respectively. After lung segmentation, the AUC of a classifier to identify bacterial or viral pneumonia increased from 0.815 to 0.879. We built a pediatric CXR dataset and exploited a deep-learning model for accurate pulmonary-thoracic segmentations. Lung segmentation can notably improve the diagnosis of bacterial or viral pneumonia.

E L ，Zhao B ，Guo Y ，Zheng C ，Zhang M ，Lin J ，Luo Y ，Cai Y ，Song X ，Liang H ... -  《-》

Standardized Interpretation of Chest Radiographs in Cases of Pediatric Pneumonia From the PERCH Study.

Chest radiographs (CXRs) are a valuable diagnostic tool in epidemiologic studies of pneumonia. The World Health Organization (WHO) methodology for the interpretation of pediatric CXRs has not been evaluated beyond its intended application as an endpoint measure for bacterial vaccine trials. The Pneumonia Etiology Research for Child Health (PERCH) study enrolled children aged 1-59 months hospitalized with WHO-defined severe and very severe pneumonia from 7 low- and middle-income countries. An interpretation process categorized each CXR into 1 of 5 conclusions: consolidation, other infiltrate, both consolidation and other infiltrate, normal, or uninterpretable. Two members of a 14-person reading panel, who had undertaken training and standardization in CXR interpretation, interpreted each CXR. Two members of an arbitration panel provided additional independent reviews of CXRs with discordant interpretations at the primary reading, blinded to previous reports. Further discordance was resolved with consensus discussion. A total of 4172 CXRs were obtained from 4232 cases. Observed agreement for detecting consolidation (with or without other infiltrate) between primary readers was 78% (κ = 0.50) and between arbitrators was 84% (κ = 0.61); agreement for primary readers and arbitrators across 5 conclusion categories was 43.5% (κ = 0.25) and 48.5% (κ = 0.32), respectively. Disagreement was most frequent between conclusions of other infiltrate and normal for both the reading panel and the arbitration panel (32% and 30% of discordant CXRs, respectively). Agreement was similar to that of previous evaluations using the WHO methodology for detecting consolidation, but poor for other infiltrates despite attempts at a rigorous standardization process.

Fancourt N ，Deloria Knoll M ，Barger-Kamate B ，de Campo J ，de Campo M ，Diallo M ，Ebruke BE ，Feikin DR ，Gleeson F ，Gong W ，Hammitt LL ，Izadnegahdar R ，Kruatrachue A ，Madhi SA ，Manduku V ，Matin FB ，Mahomed N ，Moore DP ，Mwenechanya M ，Nahar K ，Oluwalana C ，Ominde MS ，Prosperi C ，Sande J ，Suntarattiwong P ，O'Brien KL ... -  《-》

Deep Learning Method for Automated Classification of Anteroposterior and Posteroanterior Chest Radiographs.

Kim TK ，Yi PH ，Wei J ，Shin JW ，Hager G ，Hui FK ，Sair HI ，Lin CT ... -  《-》