{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,28]],"date-time":"2026-08-28T06:12:43Z","timestamp":1787897563919,"version":"build-2784847793"},"reference-count":28,"publisher":"Wiley","issue":"8","license":[{"start":{"date-parts":[[2022,2,27]],"date-time":"2022-02-27T00:00:00Z","timestamp":1645920000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["J Med Imag Rad Onc"],"published-print":{"date-parts":[[2022,12]]},"abstract":"<jats:title>Abstract<\/jats:title><jats:sec><jats:title>Introduction<\/jats:title><jats:p>The primary aim was to develop convolutional neural network (CNN)\u2010based artificial intelligence (AI) models for pneumothorax classification and segmentation for automated chest X\u2010ray (CXR) triaging. A secondary aim was to perform interpretability analysis on the best\u2010performing candidate model to determine whether the model's predictions were susceptible to bias or confounding.<\/jats:p><\/jats:sec><jats:sec><jats:title>Method<\/jats:title><jats:p>A CANDID\u2010PTX 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\u2010ROC), sensitivity and specificity, whilst segmentation performance was measured using mean Dice and true\u2010positive (TP)\u2010Dice coefficients. Interpretability analysis was performed using Grad\u2010CAM heatmaps. Finally, the best\u2010performing model was implemented for a triage simulation.<\/jats:p><\/jats:sec><jats:sec><jats:title>Results<\/jats:title><jats:p>The best\u2010performing model demonstrated a sensitivity of 0.93, specificity of 0.95 and AUC\u2010ROC of 0.94 in identifying the presence of pneumothorax. A TP\u2010Dice coefficient of 0.69 is given for segmentation performance. In triage simulation, mean reporting delay for pneumothorax\u2010containing CXRs is reduced from 9.8\u2009\u00b1\u20092\u2009days to 1.0\u2009\u00b1\u20090.5\u2009days (<jats:italic>P<\/jats:italic>\u2010value\u2009&lt;\u20090.001 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.<\/jats:p><\/jats:sec><jats:sec><jats:title>Conclusion<\/jats:title><jats:p>AI models can automate pneumothorax detection with clinically acceptable accuracy, and potentially reduce reporting delays for urgent findings when implemented as triaging tools.<\/jats:p><\/jats:sec>","DOI":"10.1111\/1754-9485.13393","type":"journal-article","created":{"date-parts":[[2022,2,28]],"date-time":"2022-02-28T07:08:54Z","timestamp":1646032134000},"page":"1035-1043","update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":34,"title":["Automated pneumothorax triaging in chest X\u2010rays in the New Zealand population using deep\u2010learning algorithms"],"prefix":"10.1111","volume":"66","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-4967-7800","authenticated-orcid":false,"given":"Sijing","family":"Feng","sequence":"first","affiliation":[{"name":"Department of Radiology Dunedin Hospital  Dunedin New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Qixiu","family":"Liu","sequence":"additional","affiliation":[{"name":"Counties Manukau Health  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Aakash","family":"Patel","sequence":"additional","affiliation":[{"name":"Dunedin School of Medicine Dunedin Hospital  Dunedin New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sibghat Ullah","family":"Bazai","sequence":"additional","affiliation":[{"name":"School of Natural and Computational Sciences Massey University  Palmerston North New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Cheng\u2010Kai","family":"Jin","sequence":"additional","affiliation":[{"name":"Auckland District Health Board  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ji Soo","family":"Kim","sequence":"additional","affiliation":[{"name":"Auckland District Health Board  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mikal","family":"Sarrafzadeh","sequence":"additional","affiliation":[{"name":"Auckland District Health Board  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Damian","family":"Azzollini","sequence":"additional","affiliation":[{"name":"Eastern Health  Melbourne Victoria Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jason","family":"Yeoh","sequence":"additional","affiliation":[{"name":"Auckland District Health Board  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Eve","family":"Kim","sequence":"additional","affiliation":[{"name":"Auckland District Health Board  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2451-456X","authenticated-orcid":false,"given":"Simon","family":"Gordon","sequence":"additional","affiliation":[{"name":"Waikato District Health Board  Hamilton New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Julian","family":"Jang\u2010Jaccard","sequence":"additional","affiliation":[{"name":"School of Natural and Computational Sciences Massey University  Palmerston North New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5792-3971","authenticated-orcid":false,"given":"Martin","family":"Urschler","sequence":"additional","affiliation":[{"name":"School of Computer Science University of Auckland  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Stuart","family":"Barnard","sequence":"additional","affiliation":[{"name":"Middlemore Hospital  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Amy","family":"Fong","sequence":"additional","affiliation":[{"name":"Department of Radiology Dunedin Hospital  Dunedin New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Cameron","family":"Simmers","sequence":"additional","affiliation":[{"name":"Department of Radiology Dunedin Hospital  Dunedin New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Gregory P","family":"Tarr","sequence":"additional","affiliation":[{"name":"Auckland City Hospital  Auckland New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ben","family":"Wilson","sequence":"additional","affiliation":[{"name":"Department of Radiology Dunedin Hospital  Dunedin New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2022,2,27]]},"reference":[{"key":"e_1_2_9_2_1","doi-asserted-by":"publisher","DOI":"10.2214\/AJR.13.11892"},{"key":"e_1_2_9_3_1","volume-title":"Clinical radiology UKworkforce census 2019 report","author":"Royal College of Radiologists","year":"2020"},{"key":"e_1_2_9_4_1","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.2018180921"},{"key":"e_1_2_9_5_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.crad.2017.11.016"},{"key":"e_1_2_9_6_1","first-page":"234","volume-title":"U\u2010net: Convolutional networks for biomedical image segmentation \u2010 International Conference on Medical Image Computing and Computer\u2010Assisted Intervention","author":"Ronneberger O","year":"2015"},{"key":"e_1_2_9_7_1","volume-title":"I\u2010Med Radiology implements world\u2010first artificial intelligence solution \u2013 Annalise CXR [Internet]","author":"I\u2010Med Radiology Network","year":"2021"},{"key":"e_1_2_9_8_1","doi-asserted-by":"crossref","unstructured":"LongJ ShelhamerE DarrellT.Fully convolutional networks for semantic segmentation. In: Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2015:3431\u20133440. [cited 2021 Jan 7].https:\/\/doi.org\/10.1109\/CVPR.2015.7298965.","DOI":"10.1109\/CVPR.2015.7298965"},{"key":"e_1_2_9_9_1","volume-title":"Very deep convolutional networks for large\u2010scale image recognition","author":"Simonyan K","year":"2015"},{"key":"e_1_2_9_10_1","doi-asserted-by":"publisher","DOI":"10.1145\/3065386"},{"key":"e_1_2_9_11_1","doi-asserted-by":"publisher","DOI":"10.1001\/jamanetworkopen.2020.0265"},{"key":"e_1_2_9_12_1","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pmed.1002686"},{"key":"e_1_2_9_13_1","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.2018182318"},{"key":"e_1_2_9_14_1","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.2019182908"},{"key":"e_1_2_9_15_1","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.2021210136"},{"key":"e_1_2_9_16_1","unstructured":"Kaggle.SIIM\u2010ACR Pneumothorax Segmentation Challenge[Internet] 2019. [cited 2020 April]. Available from URL:https:\/\/www.kaggle.com\/c\/siim\u2010acr\u2010pneumothorax\u2010segmentation."},{"key":"e_1_2_9_17_1","unstructured":"TanM LeQ.Efficientnet: Rethinking model scaling for convolutional neural networks. In Proceedings of the 36thInternational Conference on Machine Learning; 2019 Jun 9\u201315. [cited 2021 Jan 7]."},{"key":"e_1_2_9_18_1","doi-asserted-by":"crossref","unstructured":"SzegedyC IoffeS VanhouckeVet\u2009al.Inception\u2010v4 inception\u2010resnet and the impact of residual connections on learning. Proceedings of the 31stAAAI Conference on Artificial Intelligence;2017February 4\u20139; San Francisco California. Published on Aug 23 2016. Accessed on February 2 2021.","DOI":"10.1609\/aaai.v31i1.11231"},{"key":"e_1_2_9_19_1","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2017.2699184"},{"key":"e_1_2_9_20_1","unstructured":"YakubovskiyP.Segmentation Models Pytorch [Internet]. [Place unknown]: Github;2020[updated 2020; cited 2021 Dec 20]. Available from URL:https:\/\/github.com\/qubvel\/segmentation_models.pytorch."},{"key":"e_1_2_9_21_1","doi-asserted-by":"publisher","DOI":"10.3390\/info11020125"},{"key":"e_1_2_9_22_1","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-015-0816-y"},{"key":"e_1_2_9_23_1","unstructured":"KingmaDP BaJ.Adam: A method for stochastic optimization 2014. [cited 2021 Jan 7]. Available from URL:https:\/\/arxiv.org\/pdf\/1412.6980.pdf"},{"key":"e_1_2_9_24_1","doi-asserted-by":"crossref","unstructured":"GotkowskiK GonzalezC Bucher Aet\u2009al.M3d\u2010CAM: A PyTorch library to generate 3D data attention maps for medical deep learning 2020. [cited 2021 Jan 15]. Available from URL:https:\/\/arxiv.org\/pdf\/2007.00453.pdf.","DOI":"10.1007\/978-3-658-33198-6_52"},{"key":"e_1_2_9_25_1","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0242013"},{"key":"e_1_2_9_26_1","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pmed.1002697"},{"key":"e_1_2_9_27_1","first-page":"1","article-title":"On the interpretability of artificial intelligence in radiology: challenges and opportunities","volume":"2","author":"Reyes M","year":"2020","journal-title":"Radiology"},{"key":"e_1_2_9_28_1","doi-asserted-by":"crossref","unstructured":"RiberiroMT SinghS GuestrinC.\u201cWhy should I trust you?\u201d: Explaining the predictions of any classifier. KDD '16: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining;2016August 13\u201017; San Fransisco California. Published on Aug 9 2016. Accessed on December 20 2021.","DOI":"10.1145\/2939672.2939778"},{"key":"e_1_2_9_29_1","unstructured":"GoodfellowI Pouget\u2010AbadieJ MirzaMet\u2009al.Generative Adversarial Networks. Proceedings of the 27th International Conference on Neural Information Processing Systems;2014:2672\u201380."}],"container-title":["Journal of Medical Imaging and Radiation Oncology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/1754-9485.13393","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/full-xml\/10.1111\/1754-9485.13393","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/1754-9485.13393","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,8,20]],"date-time":"2023-08-20T21:17:58Z","timestamp":1692566278000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1111\/1754-9485.13393"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,2,27]]},"references-count":28,"journal-issue":{"issue":"8","published-print":{"date-parts":[[2022,12]]}},"alternative-id":["10.1111\/1754-9485.13393"],"URL":"https:\/\/doi.org\/10.1111\/1754-9485.13393","archive":["Portico"],"relation":{},"ISSN":["1754-9477","1754-9485"],"issn-type":[{"value":"1754-9477","type":"print"},{"value":"1754-9485","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,2,27]]},"assertion":[{"value":"2021-12-24","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2022-02-03","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2022-02-27","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}