{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,15]],"date-time":"2026-08-15T12:26:39Z","timestamp":1786796799392,"version":"build-2736575974"},"reference-count":35,"publisher":"SAGE Publications","license":[{"start":{"date-parts":[[2026,8,12]],"date-time":"2026-08-12T00:00:00Z","timestamp":1786492800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2026,8,12]],"date-time":"2026-08-12T00:00:00Z","timestamp":1786492800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/journals.sagepub.com\/page\/policies\/text-and-data-mining-license"}],"funder":[{"DOI":"10.13039\/100019822","name":"University of Maryland, Baltimore, Institute for Clinical & Translational Research","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100019822","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100024377","name":"Montgomery County, Maryland","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100024377","id-type":"DOI","asserted-by":"crossref"}]},{"name":"University of Maryland Strategic Partnership: MPowering the State"},{"DOI":"10.13039\/100007065","name":"NVIDIA","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100007065","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100031201","name":"Microsoft Azure","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100031201","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["journals.sagepub.com"],"crossmark-restriction":true},"short-container-title":["Can Assoc Radiol J"],"abstract":"<jats:sec>\n                    <jats:title>Purpose:<\/jats:title>\n                    <jats:p>\n                      Environmental sustainability is an emerging priority in radiology, yet the impact of deep learning training policies on greenhouse gas emissions remains poorly characterized. This study quantified the effect of training policy on carbon dioxide equivalent (CO\n                      <jats:sub>2<\/jats:sub>\n                      eq) emissions and model performance for chest radiograph classification.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods:<\/jats:title>\n                    <jats:p>\n                      Anteroposterior chest radiographs (128\u2009907 training, 24\u2009570 validation, 8282 test) were used to train 3 ImageNet-pretrained convolutional neural networks (ResNet-50, DenseNet-121, EfficientNet-B0) for 20 epochs. Three policies were evaluated: (1) retrospective optimal checkpoint selection at the validation loss minimum; (2) prospective early stopping (patience 10 epochs); and (3) fixed 20-epoch training without checkpoint selection. Per-epoch CO\n                      <jats:sub>2<\/jats:sub>\n                      eq emissions, energy, macro-averaged area under the curve (AUC), and carbon efficiency were evaluated.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results:<\/jats:title>\n                    <jats:p>\n                      Validation loss reached its minimum at median epoch 2 for ResNet-50 and DenseNet-121 and epoch 4 for EfficientNet-B0. At the retrospective optimum, macro-AUCs ranged from 0.793 to 0.800 and generated 6.2 to 7.9\u2009g CO\n                      <jats:sub>2<\/jats:sub>\n                      eq (37-46\u2009Wh) at the deployed checkpoint. However, producing this model required the full run, generating 30.8 to 49.4\u2009g CO\n                      <jats:sub>2<\/jats:sub>\n                      eq (181-291\u2009Wh) with 78% to 84% of training emissions accruing after the deployed checkpoint. Prospective early stopping had macro-AUCs equivalent to the retrospective optimum (0.793-0.800), with 31% to 38% lower total emissions (21.1-30.9 vs 30.8-49.4\u2009g CO\n                      <jats:sub>2<\/jats:sub>\n                      eq) and 57% to 76% higher carbon efficiency (25.9-37.6 vs 14.7-24.0 AUC\/kg CO\n                      <jats:sub>2<\/jats:sub>\n                      eq) compared to fixed-epoch training.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions:<\/jats:title>\n                    <jats:p>Up to 84% of total training emissions accrued after the optimal checkpoint, with relative savings dependent on the comparator. Prospective early stopping preserved performance, reduced emissions by up to 38%, and improved carbon efficiency by up to 76% versus fixed 20-epoch training.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1177\/08465371261475319","type":"journal-article","created":{"date-parts":[[2026,8,12]],"date-time":"2026-08-12T12:36:16Z","timestamp":1786538176000},"update-policy":"https:\/\/doi.org\/10.1177\/sage-journals-update-policy","source":"Crossref","is-referenced-by-count":0,"title":["Effect of Deep Learning Training Policy on Greenhouse Gas Emissions and Carbon Efficiency for Chest Radiograph Classification"],"prefix":"10.1177","author":[{"ORCID":"https:\/\/orcid.org\/0009-0007-1556-8343","authenticated-orcid":false,"given":"Nicholas","family":"Dietrich","sequence":"first","affiliation":[{"name":"Department of Medical Imaging, Temerty Faculty of Medicine, University of Toronto, ON, Canada"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David","family":"McShannon","sequence":"additional","affiliation":[{"name":"Faculty of Engineering, McMaster University, Hamilton, ON, Canada"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2289-7130","authenticated-orcid":false,"given":"Merel","family":"Huisman","sequence":"additional","affiliation":[{"name":"Department of Radiology and Nuclear Medicine, Radboud University Medical Center, Nijmegen, The Netherlands"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6519-5222","authenticated-orcid":false,"given":"Florence X.","family":"Doo","sequence":"additional","affiliation":[{"name":"University of Maryland Medical Intelligent Imaging (UM2ii) Center, Department of Radiology and Nuclear Medicine, University of Maryland, Baltimore, MD, USA"},{"name":"Center for Applied AI, University of Maryland-Institute for Health Computing (UM-IHC), North Bethesda, MD, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3077-2218","authenticated-orcid":false,"given":"Kate","family":"Hanneman","sequence":"additional","affiliation":[{"name":"Department of Medical Imaging, Temerty Faculty of Medicine, University of Toronto, ON, Canada"},{"name":"Joint Department of Medical Imaging, University Medical Imaging Toronto, University Health Network, ON, Canada"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"179","published-online":{"date-parts":[[2026,8,12]]},"reference":[{"key":"e_1_3_2_2_2","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.230229"},{"key":"e_1_3_2_3_2","doi-asserted-by":"publisher","DOI":"10.1177\/08465371251364698"},{"key":"e_1_3_2_4_2","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.232030"},{"key":"e_1_3_2_5_2","doi-asserted-by":"publisher","DOI":"10.1177\/08465371261451135"},{"key":"e_1_3_2_6_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.diii.2025.05.006"},{"key":"e_1_3_2_7_2","doi-asserted-by":"publisher","DOI":"10.1109\/3ict64318.2024.10824571"},{"key":"e_1_3_2_8_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-020-70479-z"},{"key":"e_1_3_2_9_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.jacr.2023.11.019"},{"key":"e_1_3_2_10_2","doi-asserted-by":"publisher","DOI":"10.1177\/08465371241260013"},{"key":"e_1_3_2_11_2","doi-asserted-by":"publisher","DOI":"10.1088\/2515-7620\/acf81b"},{"key":"e_1_3_2_12_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.jacr.2025.02.009"},{"key":"e_1_3_2_13_2","doi-asserted-by":"publisher","DOI":"10.1186\/s40537-021-00444-8"},{"key":"e_1_3_2_14_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.clinimag.2020.04.025"},{"key":"e_1_3_2_15_2","doi-asserted-by":"publisher","DOI":"10.7717\/peerj-cs.2968"},{"key":"e_1_3_2_16_2","doi-asserted-by":"publisher","unstructured":"Stanford Center for Artificial Intelligence in Medicine and Imaging (AIMI). CheXpert: chest X-rays. 2019. doi:10.71718\/y7pj-4v93","DOI":"10.71718\/y7pj-4v93"},{"key":"e_1_3_2_17_2","doi-asserted-by":"publisher","unstructured":"Irvin J Rajpurkar P Ko M et al. CheXpert: a large chest radiograph dataset with uncertainty labels and expert comparison. Preprint. Posted online January 21 2019. arXiv 1901.07031. doi:10.48550\/arXiv.1901.07031","DOI":"10.48550\/arXiv.1901.07031"},{"key":"e_1_3_2_18_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"e_1_3_2_19_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"e_1_3_2_20_2","doi-asserted-by":"publisher","unstructured":"Huang G Liu Z van der Maaten L Weinberger KQ. Densely connected convolutional networks. Preprint. Posted online January 28 2018. arXiv 1608.06993. doi:10.48550\/arXiv.1608.06993","DOI":"10.48550\/arXiv.1608.06993"},{"key":"e_1_3_2_21_2","doi-asserted-by":"publisher","unstructured":"Tan M Le QV. EfficientNet: rethinking model scaling for convolutional neural networks. Preprint. Posted online September 11 2020. arXiv 1905.11946. doi:10.48550\/arXiv.1905.11946","DOI":"10.48550\/arXiv.1905.11946"},{"key":"e_1_3_2_22_2","doi-asserted-by":"publisher","DOI":"10.3390\/diagnostics14161727"},{"key":"e_1_3_2_23_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.compbiolchem.2025.108586"},{"key":"e_1_3_2_24_2","doi-asserted-by":"publisher","unstructured":"Courty B Schmidt V Goyal-Kamal et al. mlco2\/codecarbon: v3.2.1. 2026. doi:10.5281\/zenodo.18129798","DOI":"10.5281\/zenodo.18129798"},{"key":"e_1_3_2_25_2","unstructured":"Government of Canada. Emission factors and reference values. 2025. Accessed March 1 2026. https:\/\/www.canada.ca\/en\/environment-climate-change\/services\/climate-change\/pricing-pollution-how-it-will-work\/output-based-pricing-system\/federal-greenhouse-gas-offset-system\/emission-factors-reference-values.html#toc4"},{"key":"e_1_3_2_26_2","unstructured":"United States Environmental Protection Agency. Greenhouse Gas Equivalencies Calculator\u2014Calculations and References. 2026. Accessed April 24 2026. https:\/\/www.epa.gov\/energy\/greenhouse-gas-equivalencies-calculator-calculations-and-references"},{"key":"e_1_3_2_27_2","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.240320"},{"key":"e_1_3_2_28_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-024-01397-x"},{"key":"e_1_3_2_29_2","doi-asserted-by":"publisher","DOI":"10.1007\/s10462-023-10457-9"},{"key":"e_1_3_2_30_2","doi-asserted-by":"publisher","DOI":"10.3233\/SHTI250585"},{"key":"e_1_3_2_31_2","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.2021210014"},{"key":"e_1_3_2_32_2","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.240588"},{"key":"e_1_3_2_33_2","doi-asserted-by":"publisher","DOI":"10.1177\/08465371251352818"},{"key":"e_1_3_2_34_2","unstructured":"Ember. Electricity Data Explorer. Ember Energy. 2025. Accessed February 14 2026. https:\/\/ember-energy.org\/data\/electricity-data-explorer\/"},{"key":"e_1_3_2_35_2","doi-asserted-by":"publisher","DOI":"10.2214\/AJR.25.33947"},{"key":"e_1_3_2_36_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.heliyon.2024.e26586"}],"container-title":["Canadian Association of Radiologists Journal"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/journals.sagepub.com\/doi\/pdf\/10.1177\/08465371261475319","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/journals.sagepub.com\/doi\/full-xml\/10.1177\/08465371261475319","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/journals.sagepub.com\/doi\/pdf\/10.1177\/08465371261475319","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,15]],"date-time":"2026-08-15T11:49:43Z","timestamp":1786794583000},"score":1,"resource":{"primary":{"URL":"https:\/\/journals.sagepub.com\/doi\/10.1177\/08465371261475319"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,12]]},"references-count":35,"alternative-id":["10.1177\/08465371261475319"],"URL":"https:\/\/doi.org\/10.1177\/08465371261475319","relation":{},"ISSN":["0846-5371","1488-2361"],"issn-type":[{"value":"0846-5371","type":"print"},{"value":"1488-2361","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,12]]},"article-number":"08465371261475319"}}