Nasreen Mahomed1,2, Bram van Ginneken3,4, Rick H H M Philipsen3,4, Jaime Melendez3, David P Moore5,6,7, Halvani Moodley8, Tanusha Sewchuran8, Denny Mathew8, Shabir A Madhi5,7. 1. Department of Radiology, Faculty of Health Sciences, University of Witwatersrand, Johannesburg, 2000, South Africa. Nasreen.Mahomed@wits.ac.za. 2. Medical Research Council, Respiratory and Meningeal Pathogens Research Unit, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa. Nasreen.Mahomed@wits.ac.za. 3. Diagnostic Image Analysis Group, Department of Radiology and Nuclear Medicine, Radboudumc, Nijmegen, the Netherlands. 4. Radboud UMC, Thirona, Nijmegen, the Netherlands. 5. Medical Research Council, Respiratory and Meningeal Pathogens Research Unit, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa. 6. Departments of Paediatrics and Child Health, Chris Hani Baragwanath Academic Hospital, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa. 7. Departments of Science/National Research Foundation: Vaccine Preventable Diseases, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa. 8. Department of Radiology, Faculty of Health Sciences, University of Witwatersrand, Johannesburg, 2000, South Africa.
Abstract
BACKGROUND: The chest radiograph is the most common imaging modality to assess childhood pneumonia. It has been used in epidemiological and vaccine efficacy/effectiveness studies on childhood pneumonia. OBJECTIVE: To develop computer-aided diagnosis (CAD4Kids) for chest radiography in children and to evaluate its accuracy in identifying World Health Organization (WHO)-defined chest radiograph primary-endpoint pneumonia compared to a consensus interpretation. MATERIALS AND METHODS: Chest radiographs were independently evaluated by three radiologists based on WHO criteria. Automatic lung field segmentation was followed by manual inspection and correction, training, feature extraction and classification. Radiographs were filtered with Gaussian derivatives on multiple scales, extracting texture features to classify each pixel in the lung region. To obtain an image score, the 95th percentile score of the pixels was used. Training and testing were done in 10-fold cross validation. RESULTS: The radiologist majority consensus reading of 858 interpretable chest radiographs included 333 (39%) categorised as primary-endpoint pneumonia, 208 (24%) as other infiltrate only and 317 (37%) as no primary-endpoint pneumonia or other infiltrate. Compared to the reference radiologist consensus reading, CAD4Kids had an area under the receiver operator characteristic (ROC) curve of 0.850 (95% confidence interval [CI] 0.823-0.876), with a sensitivity of 76% and specificity of 80% for identifying primary-endpoint pneumonia on chest radiograph. Furthermore, the ROC curve was 0.810 (95% CI 0.772-0.846) for CAD4Kids identifying primary-endpoint pneumonia compared to other infiltrate only. CONCLUSION: Further development of the CAD4Kids software and validation in multicentre studies are important for future research on computer-aided diagnosis and artificial intelligence in paediatric radiology.
BACKGROUND: The chest radiograph is the most common imaging modality to assess childhood pneumonia. It has been used in epidemiological and vaccine efficacy/effectiveness studies on childhood pneumonia. OBJECTIVE: To develop computer-aided diagnosis (CAD4Kids) for chest radiography in children and to evaluate its accuracy in identifying World Health Organization (WHO)-defined chest radiograph primary-endpoint pneumonia compared to a consensus interpretation. MATERIALS AND METHODS: Chest radiographs were independently evaluated by three radiologists based on WHO criteria. Automatic lung field segmentation was followed by manual inspection and correction, training, feature extraction and classification. Radiographs were filtered with Gaussian derivatives on multiple scales, extracting texture features to classify each pixel in the lung region. To obtain an image score, the 95th percentile score of the pixels was used. Training and testing were done in 10-fold cross validation. RESULTS: The radiologist majority consensus reading of 858 interpretable chest radiographs included 333 (39%) categorised as primary-endpoint pneumonia, 208 (24%) as other infiltrate only and 317 (37%) as no primary-endpoint pneumonia or other infiltrate. Compared to the reference radiologist consensus reading, CAD4Kids had an area under the receiver operator characteristic (ROC) curve of 0.850 (95% confidence interval [CI] 0.823-0.876), with a sensitivity of 76% and specificity of 80% for identifying primary-endpoint pneumonia on chest radiograph. Furthermore, the ROC curve was 0.810 (95% CI 0.772-0.846) for CAD4Kids identifying primary-endpoint pneumonia compared to other infiltrate only. CONCLUSION: Further development of the CAD4Kids software and validation in multicentre studies are important for future research on computer-aided diagnosis and artificial intelligence in paediatric radiology.
Entities:
Keywords:
Accuracy; Children; Computer-aided diagnosis; Pneumonia; Primary-endpoint; Radiography; World Health Organization
Authors: Thomas Cherian; E Kim Mulholland; John B Carlin; Harald Ostensen; Ruhul Amin; Margaret de Campo; David Greenberg; Rosanna Lagos; Marilla Lucero; Shabir A Madhi; Katherine L O'Brien; Steven Obaro; Mark C Steinhoff Journal: Bull World Health Organ Date: 2005-06-24 Impact factor: 9.408
Authors: Bram van Ginneken; Shigehiko Katsuragawa; Bart M ter Haar Romeny; Kunio Doi; Max A Viergever Journal: IEEE Trans Med Imaging Date: 2002-02 Impact factor: 10.048
Authors: Marianne Breuninger; Bram van Ginneken; Rick H H M Philipsen; Francis Mhimbira; Jerry J Hella; Fred Lwilla; Jan van den Hombergh; Amanda Ross; Levan Jugheli; Dirk Wagner; Klaus Reither Journal: PLoS One Date: 2014-09-05 Impact factor: 3.240
Authors: Nicholas Fancourt; Maria Deloria Knoll; Henry C Baggett; W Abdullah Brooks; Daniel R Feikin; Laura L Hammitt; Stephen R C Howie; Karen L Kotloff; Orin S Levine; Shabir A Madhi; David R Murdoch; J Anthony G Scott; Donald M Thea; Juliet O Awori; Breanna Barger-Kamate; James Chipeta; Andrea N DeLuca; Mahamadou Diallo; Amanda J Driscoll; Bernard E Ebruke; Melissa M Higdon; Yasmin Jahan; Ruth A Karron; Nasreen Mahomed; David P Moore; Kamrun Nahar; Sathapana Naorat; Micah Silaba Ominde; Daniel E Park; Christine Prosperi; Somwe Wa Somwe; Somsak Thamthitiwat; Syed M A Zaman; Scott L Zeger; Katherine L O'Brien Journal: Clin Infect Dis Date: 2017-06-15 Impact factor: 9.079