{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,13]],"date-time":"2026-03-13T09:43:18Z","timestamp":1773394998265,"version":"3.50.1"},"reference-count":44,"publisher":"PeerJ","license":[{"start":{"date-parts":[[2026,3,13]],"date-time":"2026-03-13T00:00:00Z","timestamp":1773360000000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"The Coordination for the Improvement of Higher Education Personnel"},{"DOI":"10.13039\/501100003593","name":"National Council for Scientific and Technological Development","doi-asserted-by":"crossref","award":["444564\/2024-1"],"award-info":[{"award-number":["444564\/2024-1"]}],"id":[{"id":"10.13039\/501100003593","id-type":"DOI","asserted-by":"crossref"}]},{"name":"National Institute of Science and Technology in Artificial Intelligence Applied to Smart and Sustainable Cities in the Brazilian Amazon","award":["409001\/2024-4"],"award-info":[{"award-number":["409001\/2024-4"]}]},{"name":"Municipal Fund for Sustainable Development of Cana\u00e3 dos Caraj\u00e1s","award":["001\/2023 PMCC\/UFPA\/FADESP"],"award-info":[{"award-number":["001\/2023 PMCC\/UFPA\/FADESP"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Undetected cervical lesions can progress to cancer, a leading cause of mortality among women worldwide. While automated analysis of Papanicolaou (Pap) smear images using convolutional neural networks (CNNs) has demonstrated significant potential for screening, most existing studies rely on single curated datasets. This aspect limits the understanding of model generalization to the noise and variability inherent in real-world clinical cytology.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We evaluated three CNN architectures (VGG16, ResNet50, and InceptionV3) across four curated Pap smear datasets using stratified 5-fold cross-validation. For each dataset, the model achieving the highest mean Macro-F1 score was selected for further analysis. To assess robustness against domain shift, we performed an external evaluation using a non-curated, Real-World dataset comprising routine clinical images.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>All architectures achieved robust performance on the curated benchmarks, with mean Macro-F1 scores ranging from 73.58% to 99.28%. However, performance dropped significantly when models were evaluated on the Real-World dataset (Macro-F1: 33.25\u201355.91%), highlighting the severity of the domain gap. Notably, the model trained on a combined heterogeneous dataset achieved the highest inter-domain performance, suggesting that data diversity improves robustness. Class-wise analysis revealed that high-grade lesions were most sensitive to real-world variability.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>Although CNNs achieve state-of-the-art results on curated benchmarks, their direct applicability to routine cytology workflows is hindered by domain shift. Our findings emphasize that evaluating models across heterogeneous, multi-source datasets is a prerequisite for reliable clinical deployment.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.7717\/peerj-cs.3708","type":"journal-article","created":{"date-parts":[[2026,3,13]],"date-time":"2026-03-13T08:47:08Z","timestamp":1773391628000},"page":"e3708","source":"Crossref","is-referenced-by-count":0,"title":["Assessing the efficacy of convolutional neural networks for Pap smear classification: a real world analysis"],"prefix":"10.7717","volume":"12","author":[{"given":"Sidnir Carlos Baia","family":"Ferreira","sequence":"first","affiliation":[{"name":"Graduate Program of Electrical Engineering, Universidade Federal do Par\u00e1, Bel\u00e9m, Par\u00e1, Brazil"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-4809-6945","authenticated-orcid":true,"given":"Rom\u00e1rio","family":"Silva","sequence":"additional","affiliation":[{"name":"Graduate Program of Electrical Engineering, Universidade Federal do Par\u00e1, Bel\u00e9m, Par\u00e1, Brazil"}]},{"given":"Carlos Andr\u00e9 de Mattos","family":"Teixeira","sequence":"additional","affiliation":[{"name":"Graduate Program of Electrical Engineering, Universidade Federal do Par\u00e1, Bel\u00e9m, Par\u00e1, Brazil"}]},{"given":"Josiellem","family":"Souza","sequence":"additional","affiliation":[{"name":"Oncology Research Center, Universidade Federal do Par\u00e1, Bel\u00e9m, Par\u00e1, Brazil"}]},{"given":"Evelin","family":"Gomes","sequence":"additional","affiliation":[{"name":"Department of Computing, Universidade Federal do Par\u00e1, Castanhal, Par\u00e1, Brazil"}]},{"given":"Paulo","family":"Assump\u00e7\u00e3o","sequence":"additional","affiliation":[{"name":"Oncology Research Center, Universidade Federal do Par\u00e1, Bel\u00e9m, Par\u00e1, Brazil"}]},{"given":"Jo\u00e3o","family":"Guerreiro","sequence":"additional","affiliation":[{"name":"Oncology Research Center, Universidade Federal do Par\u00e1, Bel\u00e9m, Par\u00e1, Brazil"}]},{"given":"Nandamudi","family":"Vijaykumar","sequence":"additional","affiliation":[{"name":"National Institute for Space Research, S\u00e3o Jos\u00e9 dos Campos, S\u00e3o Paulo, Brazil"}]},{"given":"Carlos Renato Lisboa","family":"Franc\u00eas","sequence":"additional","affiliation":[{"name":"Graduate Program of Electrical Engineering, Universidade Federal do Par\u00e1, Bel\u00e9m, Par\u00e1, Brazil"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3514-0401","authenticated-orcid":true,"given":"Jasmine","family":"Ara\u00fajo","sequence":"additional","affiliation":[{"name":"Graduate Program of Electrical Engineering, Universidade Federal do Par\u00e1, Bel\u00e9m, Par\u00e1, Brazil"}]}],"member":"4443","published-online":{"date-parts":[[2026,3,13]]},"reference":[{"issue":"12","key":"10.7717\/peerj-cs.3708\/ref-1","doi-asserted-by":"publisher","first-page":"2900","DOI":"10.3390\/diagnostics12122900","article-title":"Pap smear images classification using machine learning: a literature matrix","volume":"12","author":"Alias","year":"2022","journal-title":"Diagnostics"},{"issue":"11","key":"10.7717\/peerj-cs.3708\/ref-2","doi-asserted-by":"publisher","first-page":"2756","DOI":"10.3390\/diagnostics12112756","article-title":"Analysis of cytology pap smear images based on ensemble deep learning approach","volume":"12","author":"Alsalatie","year":"2022","journal-title":"Diagnostics"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-3","doi-asserted-by":"publisher","first-page":"23936","DOI":"10.1038\/s41598-025-10009-x","article-title":"CNN based method for classifying cervical cancer cells in pap smear images","volume":"15","author":"Austin","year":"2025","journal-title":"Scientific Reports"},{"issue":"Sep","key":"10.7717\/peerj-cs.3708\/ref-4","doi-asserted-by":"publisher","first-page":"1089","DOI":"10.1007\/0-387-24555-3_5","article-title":"No unbiased estimator of the variance of k-fold cross-validation","volume":"5","author":"Bengio","year":"2004","journal-title":"Journal of Machine Learning Research"},{"key":"10.7717\/peerj-cs.3708\/ref-5","first-page":"1","article-title":"Pap smear image classification using convolutional neural network","author":"Bora","year":"2016"},{"key":"10.7717\/peerj-cs.3708\/ref-6","doi-asserted-by":"crossref","first-page":"3","DOI":"10.1007\/978-3-030-33128-3_1","article-title":"Deep learning in medical image analysis","volume-title":"Deep Learning in Medical Image Analysis: Challenges and Applications","author":"Chan","year":"2020"},{"key":"10.7717\/peerj-cs.3708\/ref-7","first-page":"9268","article-title":"Class-balanced loss based on effective number of samples","author":"Cui","year":"2019"},{"issue":"7","key":"10.7717\/peerj-cs.3708\/ref-8","doi-asserted-by":"publisher","first-page":"4193","DOI":"10.3390\/ijerph19074193","article-title":"Barriers to access the pap smear test for cervical cancer screening in rural riverside populations covered by a fluvial primary healthcare team in the amazon","volume":"19","author":"Da Silva","year":"2022","journal-title":"International Journal of Environmental Research and Public Health"},{"issue":"7","key":"10.7717\/peerj-cs.3708\/ref-9","doi-asserted-by":"publisher","first-page":"111","DOI":"10.3390\/jimaging7070111","article-title":"A deep learning ensemble method to assist cytopathologists in pap test image classification","volume":"7","author":"Diniz","year":"2021","journal-title":"Journal of Imaging"},{"issue":"5","key":"10.7717\/peerj-cs.3708\/ref-10","doi-asserted-by":"publisher","first-page":"354","DOI":"10.1046\/j.0004-8666.2003.00098.x","article-title":"Women\u2019s choice in the gender and ethnicity of her obstetrician and gynaecologist","volume":"43","author":"Ekeroma","year":"2003","journal-title":"Australian and New Zealand Journal of Obstetrics and Gynaecology"},{"key":"10.7717\/peerj-cs.3708\/ref-11","volume-title":"Deep learning for vision systems","author":"Elgendy","year":"2020"},{"issue":"5","key":"10.7717\/peerj-cs.3708\/ref-12","doi-asserted-by":"publisher","first-page":"105392","DOI":"10.1016\/j.compbiomed.2022.105392","article-title":"Cervical cancer diagnosis based on modified uniform local ternary patterns and feed forward multilayer network optimized by genetic algorithm","volume":"144","author":"Fekri-Ershad","year":"2022","journal-title":"Computers in Biology and Medicine"},{"issue":"4","key":"10.7717\/peerj-cs.3708\/ref-13","doi-asserted-by":"publisher","first-page":"778","DOI":"10.1002\/ijc.33588","article-title":"Cancer statistics for the year 2020: an overview","volume":"149","author":"Ferlay","year":"2021","journal-title":"International Journal of Cancer"},{"issue":"2019","key":"10.7717\/peerj-cs.3708\/ref-14","doi-asserted-by":"publisher","first-page":"643","DOI":"10.1016\/j.future.2019.09.015","article-title":"Cervical cancer classification using convolutional neural networks and extreme learning machines","volume":"102","author":"Ghoneim","year":"2020","journal-title":"Future Generation Computer Systems"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-15","doi-asserted-by":"publisher","first-page":"3776","DOI":"10.4102\/phcfm.v15i1.3776","article-title":"Early cervical cancer screening: the influence of culture and religion","volume":"15","author":"Gutusa","year":"2023","journal-title":"African Journal of Primary Health Care & Family Medicine"},{"issue":"3","key":"10.7717\/peerj-cs.3708\/ref-16","doi-asserted-by":"publisher","first-page":"2058","DOI":"10.3892\/ol.2020.11754","article-title":"Cervical cancer in low and middle-income countries","volume":"20","author":"Hull","year":"2020","journal-title":"Oncology Letters"},{"key":"10.7717\/peerj-cs.3708\/ref-17","doi-asserted-by":"publisher","first-page":"105589","DOI":"10.17632\/zddtpgzv63.3","article-title":"Liquid based-cytology pap smear dataset for automated multi-class diagnosis of pre-cancerous and cervical cancer lesions","volume":"30","author":"Hussain","year":"2020a","journal-title":"Data in Brief"},{"key":"10.7717\/peerj-cs.3708\/ref-18","doi-asserted-by":"publisher","first-page":"101347","DOI":"10.1016\/j.tice.2020.101347","article-title":"A comprehensive study on the multi-class cervical cancer diagnostic prediction on pap smear images using a fusion-based decision from ensemble deep convolutional neural network","volume":"65","author":"Hussain","year":"2020b","journal-title":"Tissue and Cell"},{"key":"10.7717\/peerj-cs.3708\/ref-19","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.2526396","article-title":"Plotneuralnet v1.0.0. Zenodo","author":"Iqbal","year":"2018"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-20","doi-asserted-by":"publisher","first-page":"7720","DOI":"10.1038\/s41598-023-34835-z","article-title":"Computer-assisted diagnosis for an early identification of lung cancer in chest X rays","volume":"13","author":"Juan","year":"2023","journal-title":"Scientific Reports"},{"issue":"8","key":"10.7717\/peerj-cs.3708\/ref-21","doi-asserted-by":"publisher","first-page":"1838","DOI":"10.3390\/diagnostics12081838","article-title":"A comparative analysis of deep learning models for automated cross-preparation diagnosis of multi-cell liquid pap smear images","volume":"12","author":"Karasu Benyes","year":"2022","journal-title":"Diagnostics"},{"issue":"2","key":"10.7717\/peerj-cs.3708\/ref-22","doi-asserted-by":"publisher","first-page":"e1253","DOI":"10.7717\/peerj-cs.1253","article-title":"Explainability of deep learning models in medical video analysis: a survey","volume":"9","author":"Kolarik","year":"2023","journal-title":"PeerJ Computer Science"},{"issue":"3","key":"10.7717\/peerj-cs.3708\/ref-23","doi-asserted-by":"publisher","first-page":"1591","DOI":"10.1080\/03772063.2021.1997353","article-title":"Cervical cancer classification from pap smear images using modified Fuzzy C means, PCA, and KNN","volume":"68","author":"Lavanya Devi","year":"2022","journal-title":"IETE Journal of Research"},{"issue":"7","key":"10.7717\/peerj-cs.3708\/ref-24","doi-asserted-by":"publisher","first-page":"10","DOI":"10.5120\/20756-3159","article-title":"Pap smear images classification for early detection of cervical cancer","volume":"118","author":"Mbaga","year":"2015","journal-title":"International Journal of Computer Applications"},{"issue":"4","key":"10.7717\/peerj-cs.3708\/ref-25","doi-asserted-by":"publisher","first-page":"301","DOI":"10.1159\/000510991","article-title":"Challenges developing deep learning algorithms in cytology","volume":"65","author":"McAlpine","year":"2021","journal-title":"Acta Cytologica"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-26","doi-asserted-by":"publisher","first-page":"29446","DOI":"10.1038\/s41598-024-79840-y","article-title":"A lightweight deep learning method to identify different types of cervical cancer","volume":"14","author":"Mehedi","year":"2024","journal-title":"Scientific Reports"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-27","doi-asserted-by":"publisher","first-page":"1057","DOI":"10.1080\/13696998.2023.2249757","article-title":"Technical and regulatory challenges of digital health implementation in developing countries","volume":"26","author":"Meslamani","year":"2023","journal-title":"Journal of Medical Economics"},{"issue":"9770","key":"10.7717\/peerj-cs.3708\/ref-28","doi-asserted-by":"publisher","first-page":"e1490","DOI":"10.7717\/peerj-cs.1490","article-title":"Classification of Alzheimer\u2019s disease stages from magnetic resonance images using deep learning","volume":"9","author":"Mora-Rubio","year":"2023","journal-title":"PeerJ Computer Science"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-29","doi-asserted-by":"publisher","first-page":"188","DOI":"10.1007\/s10791-025-09713-z","article-title":"A two stage deep supervised learning model with inter-layer weight sharing for multi class medical image classification","volume":"28","author":"Moral","year":"2025","journal-title":"Discover Computing"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-30","doi-asserted-by":"publisher","first-page":"277","DOI":"10.13005\/bpj\/2364","article-title":"Classification of cervical cytology overlapping cell images with transfer learning architectures","volume":"15","author":"Mulmule","year":"2022","journal-title":"Biomedical and Pharmacology Journal"},{"key":"10.7717\/peerj-cs.3708\/ref-31","doi-asserted-by":"publisher","first-page":"1531817","DOI":"10.3389\/fmedt.2025.1531817","article-title":"A low-cost platform for automated cervical cytology: addressing health and socioeconomic challenges in low-resource settings","volume":"7","author":"Ocampo-L\u00f3pez-Escalera","year":"2025","journal-title":"Frontiers in Medical Technology"},{"key":"10.7717\/peerj-cs.3708\/ref-32","doi-asserted-by":"publisher","first-page":"21","DOI":"10.25259\/cmas_03_02_2021","article-title":"Cancer cervix: epidemiology and disease burden","volume":"19","author":"Pimple","year":"2022","journal-title":"Cytojournal"},{"key":"10.7717\/peerj-cs.3708\/ref-33","doi-asserted-by":"publisher","first-page":"3144","DOI":"10.1109\/ICIP.2018.8451588","article-title":"SipakMed: a new dataset for feature and image based classification of normal and pathological cervical cells in pap smear images","author":"Plissiti","year":"2018"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-34","doi-asserted-by":"publisher","first-page":"151","DOI":"10.6084\/m9.figshare.c.4960286.v2","article-title":"Cric searchable image database as a public platform for conventional pap smear cytology data","volume":"8","author":"Rezende","year":"2021","journal-title":"Scientific Data"},{"issue":"12","key":"10.7717\/peerj-cs.3708\/ref-35","doi-asserted-by":"publisher","first-page":"232","DOI":"10.20944\/preprints202410.0386.v1","article-title":"Automated cervical cancer screening using single-cell segmentation and deep learning: enhanced performance with liquid-based cytology","volume":"12","author":"Rodr\u00edguez","year":"2024","journal-title":"Computation"},{"key":"10.7717\/peerj-cs.3708\/ref-36","first-page":"618","article-title":"Grad-CAM: visual explanations from deep networks via gradient-based localization","author":"Selvaraju","year":"2017"},{"issue":"1","key":"10.7717\/peerj-cs.3708\/ref-37","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s40537-019-0197-0","article-title":"A survey on image data augmentation for deep learning","volume":"6","author":"Shorten","year":"2019","journal-title":"Journal of Big Data"},{"issue":"2","key":"10.7717\/peerj-cs.3708\/ref-38","doi-asserted-by":"publisher","first-page":"325","DOI":"10.1109\/jbhi.2020.3032060","article-title":"Measuring domain shift for deep learning in histopathology","volume":"25","author":"Stacke","year":"2020","journal-title":"IEEE Journal of Biomedical and Health Informatics"},{"issue":"10","key":"10.7717\/peerj-cs.3708\/ref-39","doi-asserted-by":"publisher","first-page":"143","DOI":"10.29322\/ijsrp.9.10.2019.p9420","article-title":"Transfer learning using VGG-16 with deep convolutional neural network for classifying images","volume":"9","author":"Tammina","year":"2019","journal-title":"International Journal of Scientific and Research Publications (IJSRP)"},{"key":"10.7717\/peerj-cs.3708\/ref-40","volume-title":"Global strategy to accelerate the elimination of cervical cancer as a public health problem","author":"WHO","year":"2020"},{"key":"10.7717\/peerj-cs.3708\/ref-41","article-title":"Cervical cancer","author":"WHO","year":"2024"},{"issue":"5","key":"10.7717\/peerj-cs.3708\/ref-42","doi-asserted-by":"publisher","first-page":"1800","DOI":"10.3390\/app10051800","article-title":"Computer-assisted screening for cervical cancer using digital image processing of pap smear images","volume":"10","author":"Win","year":"2020","journal-title":"Applied Sciences"},{"issue":"6","key":"10.7717\/peerj-cs.3708\/ref-43","doi-asserted-by":"publisher","first-page":"1633","DOI":"10.1109\/jbhi.2017.2705583","article-title":"DeepPap: deep convolutional networks for cervical cell classification","volume":"21","author":"Zhang","year":"2017","journal-title":"IEEE Journal of Biomedical and Health Informatics"},{"issue":"6","key":"10.7717\/peerj-cs.3708\/ref-44","doi-asserted-by":"publisher","first-page":"720","DOI":"10.21147\/j.issn.1000-9604.2020.06.05","article-title":"Cervical cancer: epidemiology, risk factors and screening","volume":"32","author":"Zhang","year":"2020","journal-title":"Chinese Journal of Cancer Research"}],"container-title":["PeerJ Computer Science"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/peerj.com\/articles\/cs-3708.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/peerj.com\/articles\/cs-3708.xml","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/peerj.com\/articles\/cs-3708.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/peerj.com\/articles\/cs-3708.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,13]],"date-time":"2026-03-13T08:47:14Z","timestamp":1773391634000},"score":1,"resource":{"primary":{"URL":"https:\/\/peerj.com\/articles\/cs-3708"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,13]]},"references-count":44,"alternative-id":["10.7717\/peerj-cs.3708"],"URL":"https:\/\/doi.org\/10.7717\/peerj-cs.3708","archive":["CLOCKSS","LOCKSS","Portico"],"relation":{},"ISSN":["2376-5992"],"issn-type":[{"value":"2376-5992","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3,13]]},"article-number":"e3708"}}