{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,23]],"date-time":"2026-10-23T20:50:54Z","timestamp":1792788654947,"version":"4.5.0"},"reference-count":34,"publisher":"Public Library of Science (PLoS)","issue":"10","license":[{"start":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T00:00:00Z","timestamp":1791417600000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100000038","name":"Natural Sciences and Engineering Research Council of Canada","doi-asserted-by":"publisher","award":["Discovery"],"award-info":[{"award-number":["Discovery"]}],"id":[{"id":"10.13039\/501100000038","id-type":"DOI","asserted-by":"publisher"}]},{"name":"University Health Network Foundation","id":[{"id":"https:\/\/ror.org\/042xt5161","id-type":"ROR","asserted-by":"crossref"}]}],"content-domain":{"domain":["www.plosdigitalhealth.org"],"crossmark-restriction":false},"short-container-title":["PLOS Digit Health"],"abstract":"<jats:p>Machine learning models have seen a lot of success in medical image segmentation domain. However, one of the challenges that they face are confounders or shortcuts: spurious correlations or biases in the training data that affect the resulting models. One example of such confounders for surgical machine learning is the setup of surgical equipment, including tools and lighting. Using the task of identification of safe and dangerous zones of dissection in laparoscopic cholecystectomy images and videos as a use-case, we inspect two equipment-induced biases: the location of surgical tools in the field of view and the direction of lighting, both of which are tightly related to the region of interest. We propose methods for evaluating the severity of these biases based on measurements of model consistency under real or simulated exposure to these shortcuts, and augmentation-based methods for mitigating them. We show that our tool bias mitigations based on pasting tool images in random locations improve the models\u2019 prediction consistency under tool movements by 9 percentage points in the most inconsistent cases, and by 4 percentage points on average. Our lighting bias mitigations based on simulated lighting adjustments help reduce fraction of pixels originally predicted as belonging to the dangerous zone that may flip to safe under light changes from 5% to 1.5%, without compromising segmentation quality.<\/jats:p>","DOI":"10.1371\/journal.pdig.0001421","type":"journal-article","created":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T17:53:34Z","timestamp":1791482014000},"page":"e0001421","update-policy":"https:\/\/doi.org\/10.1371\/journal.pdig.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["Analysis and mitigation of equipment-induced shortcuts in AI models for laparoscopic cholecystectomy"],"prefix":"10.1371","volume":"5","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6152-6853","authenticated-orcid":true,"given":"Sergey","family":"Protserov","sequence":"first","affiliation":[{"name":"University of Toronto","place":["Toronto, Ontario, Canada"],"department":["Department of Computer Science"]},{"name":"Vector Institute for Artificial Intelligence","place":["Toronto, Ontario, Canada"]},{"name":"University Health Network","place":["Toronto, Ontario, Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Anastasiia","family":"Repalo","sequence":"additional","affiliation":[{"name":"University of Toronto","place":["Toronto, Ontario, Canada"],"department":["Department of Computer Science"]},{"name":"Vector Institute for Artificial Intelligence","place":["Toronto, Ontario, Canada"]},{"name":"University Health Network","place":["Toronto, Ontario, Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pouria","family":"Mashouri","sequence":"additional","affiliation":[{"name":"University Health Network","place":["Toronto, Ontario, Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jaryd","family":"Hunter","sequence":"additional","affiliation":[{"name":"University Health Network","place":["Toronto, Ontario, Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Caterina","family":"Masino","sequence":"additional","affiliation":[{"name":"University Health Network","place":["Toronto, Ontario, Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Amin","family":"Madani","sequence":"additional","affiliation":[{"name":"Vector Institute for Artificial Intelligence","place":["Toronto, Ontario, Canada"]},{"name":"University Health Network","place":["Toronto, Ontario, Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7947-2243","authenticated-orcid":true,"given":"Michael","family":"Brudno","sequence":"additional","affiliation":[{"name":"University of Toronto","place":["Toronto, Ontario, Canada"],"department":["Department of Computer Science"]},{"name":"Vector Institute for Artificial Intelligence","place":["Toronto, Ontario, Canada"]},{"name":"University Health Network","place":["Toronto, Ontario, Canada"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,10,8]]},"reference":[{"issue":"4","key":"pdig.0001421.ref001","doi-asserted-by":"crossref","first-page":"460","DOI":"10.1097\/01.SLA.0000060680.92690.E9","article-title":"Causes and prevention of laparoscopic bile duct injuries: Analysis of 252 cases from a human factors and cognitive psychology perspective","volume":"237","author":"LW Way","year":"2003","journal-title":"Ann Surg"},{"issue":"2","key":"pdig.0001421.ref002","doi-asserted-by":"crossref","first-page":"255","DOI":"10.1097\/SLA.0000000000001962","article-title":"What are the principles that guide behaviors in the operating room?: Creating a framework to define and measure performance","volume":"265","author":"A Madani","year":"2017","journal-title":"Ann Surg"},{"issue":"12","key":"pdig.0001421.ref003","doi-asserted-by":"crossref","first-page":"8737","DOI":"10.1007\/s00464-022-09421-5","article-title":"Computer-aided anatomy recognition in intrathoracic and -abdominal surgery: A systematic review","volume":"36","author":"RB den Boer","year":"2022","journal-title":"Surg Endosc"},{"key":"pdig.0001421.ref004","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning.","volume":"521","author":"Y LeCun","year":"2015","journal-title":"Nature"},{"key":"pdig.0001421.ref005","unstructured":"Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, et al. Attention is All you Need. In: Neural Information Processing Systems; 2017. Available from: https:\/\/api.semanticscholar.org\/CorpusID:13756489"},{"issue":"1","key":"pdig.0001421.ref006","doi-asserted-by":"crossref","first-page":"5","DOI":"10.1038\/s41746-020-00376-2","article-title":"Deep learning-enabled medical computer vision","volume":"4","author":"A Esteva","year":"2021","journal-title":"NPJ Digit Med"},{"key":"pdig.0001421.ref007","doi-asserted-by":"crossref","DOI":"10.20517\/ais.2021.15","article-title":"Computer vision and machine learning for medical image analysis: Recent advances, challenges, and way forward","author":"E Elyan","year":"2022","journal-title":"Art Int Surg"},{"issue":"4","key":"pdig.0001421.ref008","doi-asserted-by":"crossref","first-page":"684","DOI":"10.1097\/SLA.0000000000004425","article-title":"Machine learning for surgical phase recognition: A systematic review","volume":"273","author":"CR Garrow","year":"2021","journal-title":"Ann Surg"},{"issue":"5","key":"pdig.0001421.ref009","doi-asserted-by":"crossref","first-page":"937","DOI":"10.1007\/s00264-022-05346-9","article-title":"Applications of artificial intelligence and machine learning for the hip and knee surgeon: Current state and implications for the future","volume":"46","author":"C Nich","year":"2022","journal-title":"Int Orthop"},{"issue":"6","key":"pdig.0001421.ref010","doi-asserted-by":"crossref","first-page":"456","DOI":"10.1159\/000511934","article-title":"Computer vision in the surgical operating room","volume":"36","author":"F Chadebecq","year":"2020","journal-title":"Visc Med"},{"issue":"1","key":"pdig.0001421.ref011","doi-asserted-by":"crossref","first-page":"163","DOI":"10.1038\/s41746-022-00707-5","article-title":"Computer vision in surgery: From potential to clinical value","volume":"5","author":"P Mascagni","year":"2022","journal-title":"NPJ Digit Med"},{"issue":"10","key":"pdig.0001421.ref012","doi-asserted-by":"crossref","first-page":"7412","DOI":"10.1007\/s00464-023-10335-z","article-title":"Technical skill assessment in minimally invasive surgery using artificial intelligence: A systematic review","volume":"37","author":"R Pedrett","year":"2023","journal-title":"Surg Endosc"},{"issue":"3","key":"pdig.0001421.ref013","doi-asserted-by":"crossref","first-page":"91","DOI":"10.1016\/j.lers.2023.07.001","article-title":"Current applications of artificial intelligence-based computer vision in laparoscopic surgery","volume":"6","author":"K Guo","year":"2023","journal-title":"Laparosc Endosc Robot Surg"},{"issue":"11","key":"pdig.0001421.ref014","doi-asserted-by":"crossref","first-page":"665","DOI":"10.1038\/s42256-020-00257-z","article-title":"Shortcut learning in deep neural networks","volume":"2","author":"R Geirhos","year":"2020","journal-title":"Nat Mach Intell"},{"issue":"11","key":"pdig.0001421.ref015","article-title":"Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: A cross-sectional study","volume":"15","author":"JR Zech","year":"2018","journal-title":"PLoS Med"},{"key":"pdig.0001421.ref016","doi-asserted-by":"crossref","unstructured":"Lin M, Weng N, Mikolaj K, Bashir Z, Svendsen MBS, Tolsgaard MG, et al. Shortcut Learning in Medical Image Segmentation. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. 2024. Available from: https:\/\/api.semanticscholar.org\/CorpusID:268356676","DOI":"10.1007\/978-3-031-72111-3_59"},{"issue":"2","key":"pdig.0001421.ref017","doi-asserted-by":"crossref","first-page":"363","DOI":"10.1097\/SLA.0000000000004594","article-title":"Artificial intelligence for intraoperative guidance","volume":"276","author":"A Madani","year":"2022","journal-title":"Ann Surg"},{"issue":"1","key":"pdig.0001421.ref018","doi-asserted-by":"crossref","first-page":"231","DOI":"10.1038\/s41746-024-01225-2","article-title":"Development, deployment and scaling of operating room-ready artificial intelligence for real-time surgical decision support","volume":"7","author":"S Protserov","year":"2024","journal-title":"NPJ Digit Med"},{"issue":"1","key":"pdig.0001421.ref019","doi-asserted-by":"crossref","first-page":"86","DOI":"10.1109\/TMI.2016.2593957","article-title":"EndoNet: A deep architecture for recognition tasks on laparoscopic videos","volume":"36","author":"AP Twinanda","year":"2017","journal-title":"IEEE Trans Med Imaging"},{"issue":"5","key":"pdig.0001421.ref020","doi-asserted-by":"crossref","first-page":"955","DOI":"10.1097\/SLA.0000000000004351","article-title":"Artificial intelligence for surgical safety","volume":"275","author":"P Mascagni","year":"2022","journal-title":"Ann Surg"},{"issue":"8","key":"pdig.0001421.ref021","doi-asserted-by":"crossref","first-page":"6113","DOI":"10.1007\/s00464-022-09108-x","article-title":"Surgesture: A novel instrument based on surgical actions for objective skill assessment","volume":"36","author":"Z Chen","year":"2022","journal-title":"Surg Endosc"},{"issue":"2","key":"pdig.0001421.ref022","doi-asserted-by":"crossref","first-page":"257","DOI":"10.1007\/s11548-022-02777-y","article-title":"Development and validation of a deep learning-based laparoscopic system for improving video quality","volume":"18","author":"Q Zheng","year":"2023","journal-title":"Int J Comput Assist Radiol Surg"},{"issue":"1","key":"pdig.0001421.ref023","doi-asserted-by":"crossref","first-page":"40","DOI":"10.3390\/diagnostics12010040","article-title":"Uncovering and correcting shortcut learning in machine learning models for skin cancer diagnosis","volume":"12","author":"M Nauta","year":"2021","journal-title":"Diagnostics (Basel)"},{"key":"pdig.0001421.ref024","unstructured":"Kirichenko P, Izmailov P, Wilson AG. Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations. arXiv preprint arXiv:2204.02937. 2022. Available from: https:\/\/api.semanticscholar.org\/CorpusID:247996741"},{"key":"pdig.0001421.ref025","doi-asserted-by":"crossref","unstructured":"Puli A, Zhang LH, Wald Y, Ranganath R. Don\u2019t blame Dataset Shift! Shortcut Learning due to Gradients and Cross Entropy. ArXiv. 2023. arXiv:2308.12553. Available from: https:\/\/api.semanticscholar.org\/CorpusID:261101253","DOI":"10.52202\/075280-3146"},{"key":"pdig.0001421.ref026","doi-asserted-by":"crossref","first-page":"102433","DOI":"10.1016\/j.media.2022.102433","article-title":"Rendezvous: Attention mechanisms for the recognition of surgical action triplets in endoscopic videos","volume":"78","author":"CI Nwoye","year":"2022","journal-title":"Med Image Anal"},{"key":"pdig.0001421.ref027","unstructured":"Hong WY, Kao CL, Kuo YH, Wang JR, Chang WL, Shih CS. CholecSeg8k: A Semantic Segmentation Dataset for Laparoscopic Cholecystectomy Based on Cholec80. 2020. Available from: https:\/\/arxiv.org\/abs\/2012.12453"},{"key":"pdig.0001421.ref028","article-title":"The Endoscapes dataset for surgical scene segmentation, object detection, and critical view of safety assessment: Official splits and benchmark","author":"A Murali","year":"2023","journal-title":"arXiv preprint arXiv:2312.12429"},{"issue":"1","key":"pdig.0001421.ref029","doi-asserted-by":"crossref","first-page":"825","DOI":"10.1038\/s41597-025-05163-w","article-title":"CholecInstanceSeg: A tool instance segmentation dataset for laparoscopic surgery","volume":"12","author":"O Alabi","year":"2025","journal-title":"Sci Data"},{"key":"pdig.0001421.ref030","doi-asserted-by":"crossref","unstructured":"Nwoye CI, Elgohary K, Srinivas A, Zaid F, Lavanchy JL, Padoy N. CholecTrack20: A Multi-Perspective Tracking Dataset for Surgical Tools. In: Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2025.","DOI":"10.1109\/CVPR52734.2025.00836"},{"key":"pdig.0001421.ref031","unstructured":"Yang J, Gao M, Li Z, Gao S, Wang F, Zheng F. Track Anything: Segment Anything meets videos. arXiv preprint arXiv:2304.11968. 2023. https:\/\/doi.org\/10.48550\/arXiv.2304.11968"},{"key":"pdig.0001421.ref032","unstructured":"Xie E, Wang W, Yu Z, Anandkumar A, \u00c1lvarez JM, Luo P. SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers. In: Neural Information Processing Systems. 2021. Available from: https:\/\/api.semanticscholar.org\/CorpusID:235254713"},{"key":"pdig.0001421.ref033","doi-asserted-by":"crossref","unstructured":"Ronneberger O, Fischer P, Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation. ArXiv. 2015. arXiv:1505.04597. Available from: https:\/\/api.semanticscholar.org\/CorpusID:3719281","DOI":"10.1007\/978-3-319-24574-4_28"},{"issue":"1","key":"pdig.0001421.ref034","doi-asserted-by":"crossref","first-page":"1","DOI":"10.18637\/jss.v067.i01","article-title":"Fitting linear mixed-effects models using lme4","volume":"67","author":"D Bates","year":"2015","journal-title":"J Stat Softw"}],"container-title":["PLOS Digital Health"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pdig.0001421","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T17:53:40Z","timestamp":1791482020000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pdig.0001421"}},"subtitle":[],"editor":[{"given":"Adam","family":"Hulman","sequence":"first","affiliation":[{"name":"Aarhus University Hospital","place":["DENMARK"]}],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,10,8]]},"references-count":34,"journal-issue":{"issue":"10","published-online":{"date-parts":[[2026,10,8]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pdig.0001421","relation":{},"ISSN":["2767-3170"],"issn-type":[{"value":"2767-3170","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,8]]}}}