{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,23]],"date-time":"2026-10-23T21:21:58Z","timestamp":1792790518893,"version":"4.5.0"},"reference-count":42,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2022,10,11]],"date-time":"2022-10-11T00:00:00Z","timestamp":1665446400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Radio"],"abstract":"<jats:p>As deep learning is widely used in the radiology field, the explainability of Artificial Intelligence (AI) models is becoming increasingly essential to gain clinicians\u2019 trust when using the models for diagnosis. In this research, three experiment sets were conducted with a U-Net architecture to improve the disease classification performance while enhancing the heatmaps corresponding to the model's focus through incorporating heatmap generators during training. All experiments used the dataset that contained chest radiographs, associated labels from one of the three conditions [\u201cnormal\u201d, \u201ccongestive heart failure (CHF)\u201d, and \u201cpneumonia\u201d], and numerical information regarding a radiologist's eye-gaze coordinates on the images. The paper that introduced this dataset developed a U-Net model, which was treated as the baseline model for this research, to show how the eye-gaze data can be used in multi-modal training for explainability improvement and disease classification. To compare the classification performances among this research's three experiment sets and the baseline model, the 95% confidence intervals (CI) of the area under the receiver operating characteristic curve (AUC) were measured. The best method achieved an AUC of 0.913 with a 95% CI of [0.860, 0.966]. \u201cPneumonia\u201d and \u201cCHF\u201d classes, which the baseline model struggled the most to classify, had the greatest improvements, resulting in AUCs of 0.859 with a 95% CI of [0.732, 0.957] and 0.962 with a 95% CI of [0.933, 0.989], respectively. The decoder of the U-Net for the best-performing proposed method generated heatmaps that highlight the determining image parts in model classifications. These predicted heatmaps, which can be used for the explainability of the model, also improved to align well with the radiologist's eye-gaze data. Hence, this work showed that incorporating heatmap generators and eye-gaze information into training can simultaneously improve disease classification and provide explainable visuals that align well with how the radiologist viewed the chest radiographs when making diagnosis.<\/jats:p>","DOI":"10.3389\/fradi.2022.991683","type":"journal-article","created":{"date-parts":[[2022,10,11]],"date-time":"2022-10-11T09:06:33Z","timestamp":1665479193000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":25,"title":["Improving disease classification performance and explainability of deep learning models in radiology with heatmap generators"],"prefix":"10.3389","volume":"2","author":[{"given":"Akino","family":"Watanabe","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sara","family":"Ketabi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Khashayar","family":"Namdar","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Farzad","family":"Khalvati","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2022,10,11]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"18","DOI":"10.3390\/e23010018","article-title":"Explainable AI: a review of machine learning interpretability methods","volume":"23","author":"Linardatos","year":"2021","journal-title":"Entropy"},{"key":"B2","doi-asserted-by":"publisher","first-page":"52","DOI":"10.3390\/jimaging6060052","article-title":"Explainable deep learning models in medical image analysis","volume":"6","author":"Singh","year":"2020","journal-title":"J Imaging"},{"key":"B3","doi-asserted-by":"publisher","first-page":"e1424","DOI":"10.1002\/widm.1424","article-title":"Explainable artificial intelligence: an analytical review","volume":"11","author":"Angelov","year":"2021","journal-title":"Wiley Interdiscip Rev"},{"key":"B4","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1002\/mp.15359","article-title":"A review of explainable and interpretable AI with applications in COVID-19 imaging","volume":"49","author":"Fuhrman","year":"2022","journal-title":"Med. Phys"},{"key":"B5","author":"Holzinger","year":""},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.1186\/s12911-020-01332-6","article-title":"Explainability for artificial intelligence in healthcare: a multidisciplinary perspective","volume":"310","author":"Amann","year":"2020","journal-title":"BMC Med Inform Decis Mak"},{"key":"B7","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.2020190043","article-title":"On the interpretability of artificial intelligence in radiology: challenges and opportunities","volume":"2","author":"Reyes","year":"2020","journal-title":"Radiol Artif Intell"},{"key":"B8","doi-asserted-by":"publisher","DOI":"10.1007\/s43681-022-00161-9","article-title":"Deep learning in radiology: ethics of data and on the value of algorithm transparency, interpretability and explainability","author":"Fernandez-Quilez","year":"2022","journal-title":"AI Ethics"},{"key":"B9","doi-asserted-by":"publisher","first-page":"102125","DOI":"10.1016\/j.media.2021.102125","article-title":"Deep learning for chest x-ray analysis: a survey","volume":"72","author":"\u00c7all\u0131","year":"2021","journal-title":"Med Image Anal"},{"key":"B10","author":"Shad","year":""},{"key":"B11","doi-asserted-by":"publisher","first-page":"014001","DOI":"10.1117\/1.JMI.8.S1.014001","article-title":"COVID-19 detection and heatmap generation in chest x-ray images","volume":"8","author":"Kusakunniran","year":"2021","journal-title":"J Med Imaging"},{"key":"B12","doi-asserted-by":"publisher","first-page":"92","DOI":"10.1038\/s41597-021-00863-5","article-title":"Creation and validation of a chest x-ray dataset with eye-tracking and report dictation for AI development","volume":"8","author":"Karargyris","year":"2021","journal-title":"Sci Data"},{"key":"B13","author":"Rong","year":""},{"key":"B14","doi-asserted-by":"crossref","first-page":"234","DOI":"10.1007\/978-3-319-24574-4_28","article-title":"U-net: convolutional networks for biomedical image segmentation","volume-title":"Medical image computing and computer-assisted intervention\u2014MICCAI 2015","author":"Ronneberger","year":"2015"},{"key":"B15","doi-asserted-by":"publisher","first-page":"103933","DOI":"10.1016\/j.isci.2022.103933","article-title":"Improved image classification explainability with high-accuracy heatmaps","volume":"25","author":"Preechakul","year":"2022","journal-title":"iScience"},{"key":"B16","doi-asserted-by":"publisher","first-page":"105111","DOI":"10.1016\/j.compbiomed.2021.105111","article-title":"Transparency of deep neural networks for medical image analysis: a review of interpretability methods","volume":"140","author":"Salahuddin","year":"2022","journal-title":"Comput Biol Med"},{"key":"B17","first-page":"618","author":"Selvaraju","year":"2017"},{"key":"B18","doi-asserted-by":"crossref","first-page":"818","DOI":"10.1007\/978-3-319-10590-1_53","article-title":"Visualizing and understanding convolutional networks","volume-title":"Computer vision\u2014ECCV 2014","author":"Zeiler","year":"2014"},{"key":"B19","author":"Springenberg","year":""},{"key":"B20","first-page":"2921","author":"Zhou","year":"2016"},{"key":"B21","doi-asserted-by":"publisher","first-page":"102239","DOI":"10.1016\/j.displa.2022.102239","article-title":"A review of visualization-as-explanation techniques for convolutional neural networks and their evaluation","volume":"73","author":"Mohamed","year":"2022","journal-title":"Displays"},{"key":"B22","first-page":"839","author":"Chattopadhay","year":"2018"},{"key":"B23","author":"Smilkov","year":""},{"key":"B24","first-page":"3145","author":"Shrikumar","year":"2017"},{"key":"B25","first-page":"1135","author":"Ribeiro","year":"2016"},{"key":"B26","doi-asserted-by":"publisher","first-page":"26","DOI":"10.1016\/j.media.2018.12.007","article-title":"Learning to detect chest radiographs containing pulmonary lesions using visual attention networks","volume":"53","author":"Pesce","year":"2019","journal-title":"Med Image Anal"},{"key":"B27","first-page":"249","author":"Viniavskyi","year":"2020"},{"key":"B28","author":"Chaudhry","year":"2017"},{"key":"B29","author":"Wei","year":"2017"},{"key":"B30","doi-asserted-by":"crossref","first-page":"239","DOI":"10.1007\/978-3-319-46723-8_28","article-title":"Self-transfer learning for weakly supervised lesion localization","volume-title":"Medical image computing and computer-assisted intervention\u2014mICCAI 2016","author":"Hwang","year":"2016"},{"key":"B31","author":"Rajpurkar","year":""},{"key":"B32","doi-asserted-by":"publisher","first-page":"10.1088\/1361-6560\/ac176d","DOI":"10.1088\/1361-6560\/ac176d","article-title":"Saliency-guided deep learning network for automatic tumor bed volume delineation in post-operative breast irradiation","volume":"66","author":"Kazemimoghadam","year":"2021","journal-title":"Phys Med Biol"},{"key":"B33","author":"Ismail","year":""},{"key":"B34","author":"Li","year":"2018"},{"key":"B35","author":"Wang","year":"2019"},{"key":"B36","author":"Ross","year":"2017"},{"key":"B37","year":""},{"key":"B38","author":"Tan","year":""},{"key":"B39","doi-asserted-by":"publisher","first-page":"168","DOI":"10.1016\/j.neucom.2020.08.011","article-title":"Explaining the black-box model: a survey of local interpretation methods for deep neural networks","volume":"419","author":"Liang","year":"2021","journal-title":"Neurocomputing"},{"key":"B40","doi-asserted-by":"publisher","first-page":"415","DOI":"10.1007\/s10506-020-09260-6","article-title":"Explainable AI under contract and tort law: legal incentives and technical challenges","volume":"28","author":"Hacker","year":"2020","journal-title":"Artificial Intelligence Law"},{"key":"B41","doi-asserted-by":"publisher","first-page":"665","DOI":"10.1038\/s42256-020-00257-z","article-title":"Shortcut learning in deep neural networks","volume":"2","author":"Geirhos","year":"2020","journal-title":"Nat Mach Intell"},{"key":"B42","author":"Granese","year":""}],"container-title":["Frontiers in Radiology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fradi.2022.991683\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,10,11]],"date-time":"2022-10-11T09:06:39Z","timestamp":1665479199000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fradi.2022.991683\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,10,11]]},"references-count":42,"alternative-id":["10.3389\/fradi.2022.991683"],"URL":"https:\/\/doi.org\/10.3389\/fradi.2022.991683","relation":{},"ISSN":["2673-8740"],"issn-type":[{"value":"2673-8740","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,10,11]]},"article-number":"991683"}}