{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T23:38:55Z","timestamp":1768001935993,"version":"3.49.0"},"reference-count":49,"publisher":"Wiley","license":[{"start":{"date-parts":[[2021,6,13]],"date-time":"2021-06-13T00:00:00Z","timestamp":1623542400000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Progetti di Ricerca Sapienza"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Computational and Mathematical Methods in Medicine"],"published-print":{"date-parts":[[2021,6,13]]},"abstract":"<jats:p>Deep convolutional networks have become a powerful tool for medical imaging diagnostic. In pathology, most efforts have been focused in the subfield of histology, while cytopathology (which studies diagnostic tools at the cellular level) remains underexplored. In this paper, we propose a novel deep learning model for cancer detection from urinary cytopathology screening images. We leverage recent ideas from the field of multioutput neural networks to provide a model that can efficiently train even on small-scale datasets, such as those typically found in real-world scenarios. Additionally, we argue that calibration (i.e., providing confidence levels that are aligned with the ground truth probability of an event) has been a major shortcoming of prior works, and we experiment a number of techniques to provide a well-calibrated model. We evaluate the proposed algorithm on a novel dataset, and we show that the combination of focal loss, multiple outputs, and temperature scaling provides a model that is significantly more accurate and calibrated than a baseline deep convolutional network.<\/jats:p>","DOI":"10.1155\/2021\/5569458","type":"journal-article","created":{"date-parts":[[2021,6,14]],"date-time":"2021-06-14T19:05:08Z","timestamp":1623697508000},"page":"1-11","source":"Crossref","is-referenced-by-count":13,"title":["A Calibrated Multiexit Neural Network for Detecting Urothelial Cancer Cells"],"prefix":"10.1155","volume":"2021","author":[{"given":"L.","family":"Lilli","sequence":"first","affiliation":[{"name":"Department of Information Engineering, Electronics and Telecommunications (DIET), Sapienza University of Rome, Italy"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2046-5505","authenticated-orcid":true,"given":"E.","family":"Giarnieri","sequence":"additional","affiliation":[{"name":"Faculty of Medicine and Psychology, Sapienza University of Rome, Italy"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0881-8344","authenticated-orcid":true,"given":"S.","family":"Scardapane","sequence":"additional","affiliation":[{"name":"Department of Information Engineering, Electronics and Telecommunications (DIET), Sapienza University of Rome, Italy"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"311","reference":[{"key":"1","doi-asserted-by":"publisher","DOI":"10.1038\/nrdp.2017.23"},{"key":"2","volume-title":"Urologic Surgical Pathology","author":"D. G. Bostwick","year":"2008"},{"key":"3","doi-asserted-by":"publisher","DOI":"10.1145\/1102351.1102430"},{"key":"4","doi-asserted-by":"publisher","DOI":"10.1159\/000510474"},{"key":"5","doi-asserted-by":"publisher","DOI":"10.1111\/cyt.12345"},{"key":"6","doi-asserted-by":"publisher","DOI":"10.1159\/000446270"},{"key":"7","doi-asserted-by":"publisher","DOI":"10.1002\/cncy.21999"},{"key":"8","doi-asserted-by":"publisher","DOI":"10.1097\/00005392-199805000-00057"},{"key":"9","doi-asserted-by":"publisher","DOI":"10.1155\/2019\/2717454"},{"key":"10","doi-asserted-by":"publisher","DOI":"10.1001\/jama.2016.17216"},{"key":"11","doi-asserted-by":"publisher","DOI":"10.1038\/nature21056"},{"key":"12","doi-asserted-by":"publisher","DOI":"10.1016\/S1470-2045(19)30154-8"},{"key":"13","doi-asserted-by":"publisher","DOI":"10.1155\/2020\/9689821"},{"key":"14","first-page":"1321","article-title":"On calibration of modern neural networks","author":"C. Guo"},{"key":"15","doi-asserted-by":"publisher","DOI":"10.1136\/amiajnl-2011-000291"},{"key":"16","article-title":"Calibrating deep neural networks using focal loss","author":"J. Mukhoti","year":"2020"},{"key":"17","article-title":"Very deep convolutional networks for large-scale image recognition","author":"K. Simonyan","year":"2014"},{"key":"18","first-page":"234","article-title":"U-net: convolutional networks for biomedical image segmentation","author":"O. Ronneberger","year":"2015"},{"key":"19","first-page":"158","article-title":"Deep neural networks for fast segmentation of 3d medical images","author":"K. Fritscher","year":"2016"},{"key":"20","doi-asserted-by":"crossref","unstructured":"YueL.TianD.ChenW.HanX.YinM.Deep learning for heterogeneous medical data analysis2020World Wide Web","DOI":"10.1007\/s11280-019-00764-z"},{"key":"21","doi-asserted-by":"publisher","DOI":"10.4103\/2153-3539.186902"},{"key":"22","doi-asserted-by":"publisher","DOI":"10.4137\/BECB.S31601"},{"key":"23","doi-asserted-by":"publisher","DOI":"10.1016\/j.jasc.2019.03.003"},{"key":"24","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2020.2983186"},{"key":"25","doi-asserted-by":"publisher","DOI":"10.1002\/dc.23244"},{"key":"26","doi-asserted-by":"publisher","DOI":"10.1002\/cncy.22176"},{"key":"27","first-page":"609","article-title":"Obtaining calibrated probability estimates from decision trees and naive bayesian classifiers","author":"B. Zadrozny","year":"2001"},{"key":"28","doi-asserted-by":"publisher","DOI":"10.1145\/775047.775151"},{"issue":"3","key":"29","first-page":"61","article-title":"Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods","volume":"10","author":"J. Platt","year":"1999","journal-title":"Advances in Large Margin Classifiers"},{"key":"30","article-title":"Distilling the knowledge in a neural network","author":"G. Hinton","year":"2015"},{"key":"31","article-title":"Mixmatch: a holistic approach to semi-supervised learning","author":"D. Berthelot","year":"2019"},{"key":"32","doi-asserted-by":"publisher","DOI":"10.1109\/iccv.2017.324"},{"key":"33","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2015.7298594"},{"key":"34","first-page":"562","article-title":"Deeply-supervised nets","volume-title":"Artificial Intelligence and Statistics","author":"C.-Y. Lee","year":"2015"},{"key":"35","doi-asserted-by":"publisher","DOI":"10.1109\/icpr.2016.7900006"},{"key":"36","article-title":"Idk cascades: fast deep learning by learning not to overthink","author":"X. Wang","year":"2017"},{"key":"37","first-page":"527","article-title":"Adaptive neural networks for efficient inference","author":"T. Bolukbasi"},{"key":"38","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2018.2805098"},{"key":"39","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v33i01.33013812"},{"key":"40","article-title":"Forward thinking: building and training neural networks one layer at a time","author":"C. Hettinger","year":"2017"},{"key":"41","first-page":"3301","article-title":"Shallow-deep networks: understanding and mitigating network overthinking","author":"Y. Kaya"},{"key":"42","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2021.3062227"},{"key":"43","doi-asserted-by":"publisher","DOI":"10.1007\/s12559-020-09734-4"},{"key":"44","doi-asserted-by":"publisher","DOI":"10.1145\/3007192"},{"key":"45","doi-asserted-by":"publisher","DOI":"10.1016\/j.ins.2020.02.041"},{"key":"46","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2016.90"},{"key":"47","doi-asserted-by":"publisher","DOI":"10.1042\/bsr20180289"},{"key":"48","doi-asserted-by":"crossref","first-page":"135","DOI":"10.1007\/978-981-13-3600-3_13","article-title":"White blood cell classification using convolutional neural network","volume-title":"Soft Computing and Signal Processing","author":"M. Sharma","year":"2019"},{"key":"49","article-title":"An image is worth 16x16 words: transformers for image recognition at scale","author":"A. Dosovitskiy","year":"2020"}],"container-title":["Computational and Mathematical Methods in Medicine"],"original-title":[],"language":"en","link":[{"URL":"http:\/\/downloads.hindawi.com\/journals\/cmmm\/2021\/5569458.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"http:\/\/downloads.hindawi.com\/journals\/cmmm\/2021\/5569458.xml","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"http:\/\/downloads.hindawi.com\/journals\/cmmm\/2021\/5569458.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2021,6,14]],"date-time":"2021-06-14T19:05:21Z","timestamp":1623697521000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.hindawi.com\/journals\/cmmm\/2021\/5569458\/"}},"subtitle":[],"editor":[{"given":"Nadia A.","family":"Chuzhanova","sequence":"additional","affiliation":[],"role":[{"role":"editor","vocabulary":"crossref"}]}],"short-title":[],"issued":{"date-parts":[[2021,6,13]]},"references-count":49,"alternative-id":["5569458","5569458"],"URL":"https:\/\/doi.org\/10.1155\/2021\/5569458","relation":{},"ISSN":["1748-6718","1748-670X"],"issn-type":[{"value":"1748-6718","type":"electronic"},{"value":"1748-670X","type":"print"}],"subject":[],"published":{"date-parts":[[2021,6,13]]}}}