{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,22]],"date-time":"2026-10-22T05:47:25Z","timestamp":1792648045696,"version":"4.5.0"},"reference-count":36,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2023,8,11]],"date-time":"2023-08-11T00:00:00Z","timestamp":1691712000000},"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. Endocrinol."],"abstract":"<jats:sec><jats:title>Objectives<\/jats:title><jats:p>The aim of this study was to improve the diagnostic performance of nuclear medicine physicians using a deep convolutional neural network (DCNN) model and validate the results with two multicenter datasets for thyroid disease by analyzing clinical single-photon emission computed tomography (SPECT) image data.<\/jats:p><\/jats:sec><jats:sec><jats:title>Methods<\/jats:title><jats:p>In this multicenter retrospective study, 3194 SPECT thyroid images were collected for model training (n=2067), internal validation (n=514) and external validation (n=613). First, four pretrained DCNN models (AlexNet, ShuffleNetV2, MobileNetV3 and ResNet-34) for were tested multiple medical image classification of thyroid disease types (i.e., Graves\u2019 disease, subacute thyroiditis, thyroid tumor and normal thyroid). The best performing model was then subjected to fivefold cross-validation to further assess its performance, and the diagnostic performance of this model was compared with that of junior and senior nuclear medicine physicians. Finally, class-specific attentional regions were visualized with attention heatmaps using gradient-weighted class activation mapping.<\/jats:p><\/jats:sec><jats:sec><jats:title>Results<\/jats:title><jats:p>Each of the four pretrained neural networks attained an overall accuracy of more than 0.85 for the classification of SPECT thyroid images. The improved ResNet-34 model performed best, with an accuracy of 0.944. For the internal validation set, the ResNet-34 model showed higher accuracy (<jats:italic>p<\/jats:italic> &amp;lt; 0.001) when compared to that of the senior nuclear medicine physician, with an improvement of nearly 10%. Our model achieved an overall accuracy of 0.931 for the external dataset, a significantly higher accuracy than that of the senior physician (0.931 vs. 0.868, <jats:italic>p<\/jats:italic> &amp;lt; 0.001).<\/jats:p><\/jats:sec><jats:sec><jats:title>Conclusion<\/jats:title><jats:p>The DCNN-based model performed well in terms of diagnosing thyroid scintillation images. The DCNN model showed higher sensitivity and greater specificity in identifying Graves\u2019 disease, subacute thyroiditis, and thyroid tumors compared to those of nuclear medicine physicians, illustrating the feasibility of deep learning models to improve the diagnostic efficiency for assisting clinicians.<\/jats:p><\/jats:sec>","DOI":"10.3389\/fendo.2023.1224191","type":"journal-article","created":{"date-parts":[[2023,8,11]],"date-time":"2023-08-11T17:50:04Z","timestamp":1691776204000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":30,"title":["Diagnosis of thyroid disease using deep convolutional neural network models applied to thyroid scintigraphy images: a multicenter study"],"prefix":"10.3389","volume":"14","author":[{"given":"Huayi","family":"Zhao","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chenxi","family":"Zheng","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Huihui","family":"Zhang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Maohua","family":"Rao","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yixuan","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Danzhou","family":"Fang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jiahui","family":"Huang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wenqian","family":"Zhang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Gengbiao","family":"Yuan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2023,8,11]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s12940-021-00818-8","article-title":"The risk of perchlorate and iodine on the incidence of thyroid tumors and nodular goiter: A case-control study in Southeastern China","volume":"21","author":"Wang","year":"2022","journal-title":"Environ Health"},{"key":"B2","doi-asserted-by":"publisher","DOI":"10.1089\/thy.2019.0067","article-title":"Efficacy and safety of long-term universal salt iodization on thyroid disorders: epidemiological evidence from 31 provinces of mainland China","volume":"30","author":"Li","year":"2020","journal-title":"Thyroid"},{"key":"B3","doi-asserted-by":"publisher","DOI":"10.3389\/fcell.2021.740267","article-title":"Prioritizing susceptible genes for thyroid cancer based on gene interaction network","volume":"9","author":"Zhong","year":"2021","journal-title":"Front Cell Dev Biol"},{"key":"B4","doi-asserted-by":"publisher","DOI":"10.1016\/j.nwh.2015.12.002","article-title":"Hypothyroidism in women","volume":"20","author":"Dunn","year":"2016","journal-title":"Nurs women's Health"},{"key":"B5","doi-asserted-by":"publisher","DOI":"10.1007\/s00259-002-0811-8","article-title":"The continuing importance of thyroid scintigraphy in the era of high-resolution ultrasound","volume":"29","author":"Meller","year":"2002","journal-title":"Eur J Nucl Med Mol Imaging"},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.1111\/cen.14379","article-title":"Thyroid scintigraphy in the era of fine-needle aspiration cytology","volume":"94","author":"Pirola","year":"2020","journal-title":"Clin Endocrinol"},{"key":"B7","doi-asserted-by":"publisher","DOI":"10.1016\/j.sjbs.2020.05.015","article-title":"Scintigraphy has the potential to replace thyroid stimulating hormone and ultrasonography in hyperthyroidism diagnosis","volume":"27","author":"Almohammed","year":"2020","journal-title":"Saudi J Biol Sci"},{"key":"B8","doi-asserted-by":"publisher","DOI":"10.3390\/diagnostics11030557","article-title":"Long-term follow-up of thyroid incidentalomas visualized with 18f-fluorodeoxyglucose positron emission tomography-impact of thyroid scintigraphy in the diagnostic work-up","volume":"11","author":"Korsholm","year":"2021","journal-title":"Diagnostics (Basel)"},{"key":"B9","doi-asserted-by":"publisher","first-page":"PO","DOI":"10.1158\/1557-3265.ADI21-PO-056","article-title":"Abstract po-056: Importance of artificial intelligence, machine learning deep learning in the field of medicine on the future role of the physician","volume":"27","author":"Kumar","year":"2021","journal-title":"Clin Cancer Res"},{"key":"B10","doi-asserted-by":"publisher","DOI":"10.1016\/j.semcancer.2021.04.013","article-title":"Artificial intelligence in oncology: From bench to clinic","volume":"84","author":"Elkhader","year":"2022","journal-title":"Semin Cancer Biol"},{"key":"B11","doi-asserted-by":"publisher","DOI":"10.1002\/cac2.12012","article-title":"Emerging role of deep learning-based artificial intelligence in tumor pathology","volume":"40","author":"Jiang","year":"2020","journal-title":"Cancer Commun (Lond)"},{"key":"B12","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-021-01343-4","article-title":"Deep learning in histopathology: The path to the clinic","volume":"27","author":"Van der Laak","year":"2021","journal-title":"Nat Med"},{"key":"B13","doi-asserted-by":"publisher","first-page":"29S","DOI":"10.2967\/jnumed.118.220590","article-title":"Artificial intelligence in nuclear medicine","volume":"60","author":"Nensa","year":"2019","journal-title":"J Nucl Med"},{"key":"B14","doi-asserted-by":"publisher","DOI":"10.1016\/j.jid.2020.02.026","article-title":"Artificial intelligence in dermatology: A primer","volume":"140","author":"Young","year":"2020","journal-title":"J Invest Dermatol"},{"key":"B15","doi-asserted-by":"publisher","DOI":"10.1016\/j.cmpb.2022.106823","article-title":"Deep convolutional neural networks in thyroid disease detection: A multi-classification comparison by ultrasonography and computed tomography","volume":"220","author":"Zhang","year":"2022","journal-title":"Comput Methods Programs BioMed"},{"key":"B16","doi-asserted-by":"publisher","first-page":"64","DOI":"10.1007\/s00330-022-08978-y","article-title":"A concurrent, deep learning-based computer-aided detection system for prostate multiparametric mri: A performance study involving experienced and less-experienced radiologists","volume":"33","author":"Labus","year":"2023","journal-title":"Eur Radiol"},{"key":"B17","doi-asserted-by":"publisher","first-page":"193","DOI":"10.1016\/S1470-2045(18)30762-9","article-title":"Diagnosis of thyroid cancer using deep convolutional neural network models applied to sonographic images: A retrospective, multicohort, diagnostic study","volume":"20","author":"Li","year":"2019","journal-title":"Lancet Oncol"},{"key":"B18","doi-asserted-by":"publisher","DOI":"10.1016\/j.eclinm.2023.101905","article-title":"Ultrasound image-based deep learning to assist in diagnosing gross extrathyroidal extension thyroid cancer: A retrospective multicenter study","volume":"58","author":"Qi","year":"2023","journal-title":"EClinicalMedicine"},{"key":"B19","doi-asserted-by":"publisher","DOI":"10.1016\/S1470-2045(19)30333-X","article-title":"Comparison of the accuracy of human readers versus machine-learning algorithms for pigmented skin lesion classification: an open, web-based, international, diagnostic study","volume":"20","author":"Tschandl","year":"2019","journal-title":"Lancet Oncol"},{"key":"B20","doi-asserted-by":"publisher","first-page":"84","DOI":"10.1145\/3065386","article-title":"Imagenet classification with deep convolutional neural networks","volume":"60","author":"Krizhevsky","year":"2017","journal-title":"Commun ACM"},{"key":"B21","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2307.11164","article-title":"Shufflenet V2: Practical guidelines for efficient cnn architecture design","author":"Ma","year":"2018","journal-title":"Comput Vision \u2013 ECCV 2018"},{"key":"B22","doi-asserted-by":"crossref","DOI":"10.1109\/ICCV.2019.00140","article-title":"Searching for mobilenetv3","author":"Howard","year":"2019"},{"key":"B23","doi-asserted-by":"crossref","DOI":"10.1109\/CVPR.2016.90","article-title":"Deep residual learning for image recognition","author":"He","year":"2016"},{"key":"B24","doi-asserted-by":"crossref","DOI":"10.1109\/MIPR.2018.00032","article-title":"The impact of multi-optimizers and data augmentation on tensorflow convolutional neural network performance","author":"Taqi","year":"2018"},{"key":"B25","article-title":"Dropout: A simple way to prevent neural networks from overfitting","volume":"15","author":"Srivastava","year":"2014","journal-title":"J Mach Learn Res"},{"key":"B26","doi-asserted-by":"publisher","DOI":"10.1023\/A:1010091220143","article-title":"The cross-entropy method for combinatorial and continuous optimization","volume":"1","author":"Rubinstein","year":"1999","journal-title":"Method Computing Appl Probability"},{"key":"B27","doi-asserted-by":"publisher","DOI":"10.1038\/538020a","article-title":"Can we open the black box of ai","volume":"538","author":"Castelvecchi","year":"2016","journal-title":"Nat News"},{"key":"B28","doi-asserted-by":"publisher","DOI":"10.1007\/s00259-021-05569-9","article-title":"Explainable ai to improve acceptance of convolutional neural networks for automatic classification of dopamine transporter spect in the diagnosis of clinically uncertain parkinsonian syndromes","volume":"49","author":"Nazari","year":"2022","journal-title":"Eur J Nucl Med Mol Imaging"},{"key":"B29","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-019-01228-7","article-title":"Grad-cam: Visual explanations from deep networks via gradient-based localization","volume":"128","author":"Selvaraju","year":"2020","journal-title":"Int J Comput Vision"},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.1210\/clinem\/dgab870","article-title":"Convolutional neural network-based computer-assisted diagnosis of hashimoto's thyroiditis on ultrasound","volume":"107","author":"Zhao","year":"2022","journal-title":"J Clin Endocrinol Metab"},{"key":"B31","doi-asserted-by":"publisher","first-page":"19","DOI":"10.1148\/radiol.2019190372","article-title":"Lymph node metastasis prediction from primary breast cancer us images using deep learning","volume":"294","author":"Zhou","year":"2020","journal-title":"Radiology"},{"key":"B32","doi-asserted-by":"publisher","first-page":"1044","DOI":"10.1053\/j.gastro.2019.06.025","article-title":"Gastroenterologist-level identification of small-bowel diseases and normal variants by capsule endoscopy using a deep-learning model","volume":"157","author":"Ding","year":"2019","journal-title":"Gastroenterology"},{"key":"B33","doi-asserted-by":"publisher","DOI":"10.1093\/annonc\/mdy166","article-title":"Man against machine: Diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists","volume":"29","author":"Haenssle","year":"2018","journal-title":"Ann Oncol"},{"key":"B34","doi-asserted-by":"publisher","DOI":"10.1177\/0300060520982842","article-title":"Deep learning for intelligent diagnosis in thyroid scintigraphy","volume":"49","author":"Qiao","year":"2021","journal-title":"J Int Med Res"},{"key":"B35","doi-asserted-by":"publisher","DOI":"10.1155\/2019\/6212759","article-title":"Thyroid diagnosis from spect images using convolutional neural network with optimization","volume":"2019","author":"Ma","year":"2019","journal-title":"Comput Intell Neurosci"},{"key":"B36","doi-asserted-by":"publisher","DOI":"10.1007\/s11063-011-9207-8","article-title":"Reverse engineering the neural networks for rule extraction in classification problems","volume":"35","author":"Augasta","year":"2012","journal-title":"Neural Process Lett"}],"container-title":["Frontiers in Endocrinology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fendo.2023.1224191\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,8,11]],"date-time":"2023-08-11T17:50:14Z","timestamp":1691776214000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fendo.2023.1224191\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,8,11]]},"references-count":36,"alternative-id":["10.3389\/fendo.2023.1224191"],"URL":"https:\/\/doi.org\/10.3389\/fendo.2023.1224191","relation":{},"ISSN":["1664-2392"],"issn-type":[{"value":"1664-2392","type":"electronic"}],"subject":[],"published":{"date-parts":[[2023,8,11]]},"article-number":"1224191"}}