{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,8]],"date-time":"2026-04-08T11:45:21Z","timestamp":1775648721105,"version":"3.50.1"},"reference-count":54,"publisher":"Public Library of Science (PLoS)","issue":"4","license":[{"start":{"date-parts":[[2022,4,14]],"date-time":"2022-04-14T00:00:00Z","timestamp":1649894400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Agence Nationale de la Recherche-ERA-CAPS","award":["17-CAPS-0006-01"],"award-info":[{"award-number":["17-CAPS-0006-01"]}]}],"content-domain":{"domain":["www.ploscompbiol.org"],"crossmark-restriction":false},"short-container-title":["PLoS Comput Biol"],"abstract":"<jats:p>Segmenting three-dimensional (3D) microscopy images is essential for understanding phenomena like morphogenesis, cell division, cellular growth, and genetic expression patterns. Recently, deep learning (DL) pipelines have been developed, which claim to provide high accuracy segmentation of cellular images and are increasingly considered as the state of the art for image segmentation problems. However, it remains difficult to define their relative performances as the concurrent diversity and lack of uniform evaluation strategies makes it difficult to know how their results compare. In this paper, we first made an inventory of the available DL methods for 3D cell segmentation. We next implemented and quantitatively compared a number of representative DL pipelines, alongside a highly efficient non-DL method named MARS. The DL methods were trained on a common dataset of 3D cellular confocal microscopy images. Their segmentation accuracies were also tested in the presence of different image artifacts. A specific method for segmentation quality evaluation was adopted, which isolates segmentation errors due to under- or oversegmentation. This is complemented with a 3D visualization strategy for interactive exploration of segmentation quality. Our analysis shows that the DL pipelines have different levels of accuracy. Two of them, which are end-to-end 3D and were originally designed for cell boundary detection, show high performance and offer clear advantages in terms of adaptability to new data.<\/jats:p>","DOI":"10.1371\/journal.pcbi.1009879","type":"journal-article","created":{"date-parts":[[2022,4,14]],"date-time":"2022-04-14T13:26:37Z","timestamp":1649942797000},"page":"e1009879","update-policy":"https:\/\/doi.org\/10.1371\/journal.pcbi.corrections_policy","source":"Crossref","is-referenced-by-count":33,"title":["Benchmarking of deep learning algorithms for 3D instance segmentation of confocal image datasets"],"prefix":"10.1371","volume":"18","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-2543-1697","authenticated-orcid":true,"given":"Anuradha","family":"Kar","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3211-0066","authenticated-orcid":true,"given":"Manuel","family":"Petit","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6136-608X","authenticated-orcid":true,"given":"Yassin","family":"Refahi","sequence":"additional","affiliation":[]},{"given":"Guillaume","family":"Cerutti","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1202-8460","authenticated-orcid":true,"given":"Christophe","family":"Godin","sequence":"additional","affiliation":[]},{"given":"Jan","family":"Traas","sequence":"additional","affiliation":[]}],"member":"340","published-online":{"date-parts":[[2022,4,14]]},"reference":[{"key":"pcbi.1009879.ref001","first-page":"1","volume-title":"2017 International Conference on Circuit, Power and Computing Technologies (ICCPCT).","author":"RM Thomas","year":"2017"},{"key":"pcbi.1009879.ref002","doi-asserted-by":"crossref","first-page":"171","DOI":"10.1007\/s13735-020-00195-x","article-title":"A survey on instance segmentation: state of the art","volume":"9","author":"AM Hafiz","year":"2020","journal-title":"Int J Multimed Inf Retr"},{"key":"pcbi.1009879.ref003","doi-asserted-by":"crossref","first-page":"360","DOI":"10.1186\/s12859-019-2880-8","article-title":"Cell segmentation methods for label-free contrast microscopy: review and comprehensive comparison","volume":"20","author":"T Vicar","year":"2019","journal-title":"BMC Bioinformatics"},{"key":"pcbi.1009879.ref004","article-title":"Medical Image Segmentation Using Deep Learning: A Survey","author":"T Lei","year":"2020","journal-title":"arXiv"},{"key":"pcbi.1009879.ref005","doi-asserted-by":"crossref","first-page":"53","DOI":"10.3389\/fbioe.2019.00053","article-title":"Methods for segmentation and classification of digital microscopy tissue images","volume":"7","author":"QD Vu","year":"2019","journal-title":"Front Bioeng Biotechnol"},{"key":"pcbi.1009879.ref006","first-page":"1","volume-title":"OCEANS 2015\u2014Genova.","author":"F Kallasi","year":"2015"},{"key":"pcbi.1009879.ref007","doi-asserted-by":"crossref","first-page":"1277","DOI":"10.1016\/0031-3203(93)90135-J","article-title":"A review on image segmentation techniques","volume":"26","author":"NR Pal","year":"1993","journal-title":"Pattern Recogn"},{"key":"pcbi.1009879.ref008","article-title":"Globally Optimal Image Partitioning by Multicuts","author":"JH Kappes","year":"2011","journal-title":"EMMCVPR"},{"key":"pcbi.1009879.ref009","article-title":"Segmenting Planar Superpixel Adjacency Graphs w.r.t","author":"B Andres","year":"2013","journal-title":"Non-planar Superpixel Affinity Graphs. EMMCVPR"},{"key":"pcbi.1009879.ref010","doi-asserted-by":"crossref","first-page":"101","DOI":"10.1038\/nmeth.4151","article-title":"Multicut brings automated neurite segmentation closer to human performance","volume":"14","author":"T Beier","year":"2017","journal-title":"Nat Methods"},{"key":"pcbi.1009879.ref011","doi-asserted-by":"crossref","first-page":"117","DOI":"10.1016\/j.sigpro.2013.10.008","article-title":"Active contour model driven by linear speed function for local segmentation with robust initialization and applications in MR brain images","volume":"97","author":"Q Zheng","year":"2014","journal-title":"Signal Process"},{"key":"pcbi.1009879.ref012","article-title":"A Simple Method to improve Initialization Robustness for Active Contours driven by Local Region Fitting Energy","volume":"10437","author":"K. Ding","year":"2018","journal-title":"CoRR"},{"key":"pcbi.1009879.ref013","doi-asserted-by":"crossref","first-page":"e1005177","DOI":"10.1371\/journal.pcbi.1005177","article-title":"Deep Learning Automates the Quantitative Analysis of Individual Cells in Live-Cell Imaging Experiments","volume":"12","author":"DA Van Valen","year":"2016","journal-title":"PLoS Comput Biol"},{"key":"pcbi.1009879.ref014","doi-asserted-by":"crossref","first-page":"133","DOI":"10.1016\/j.compbiomed.2019.04.006","article-title":"Learn to segment single cells with deep distance estimator and deep cell detector","volume":"108","author":"W Wang","year":"2019","journal-title":"Comput Biol Med"},{"key":"pcbi.1009879.ref015","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":"pcbi.1009879.ref016","doi-asserted-by":"crossref","first-page":"2555","DOI":"10.1093\/bioinformatics\/btx188","article-title":"DeepEM3D: approaching human-level performance on 3D anisotropic EM image segmentation","volume":"33","author":"T Zeng","year":"2017","journal-title":"Bioinformatics"},{"key":"pcbi.1009879.ref017","doi-asserted-by":"crossref","first-page":"952","DOI":"10.1002\/cyto.a.23863","article-title":"Evaluation of deep learning strategies for nucleus segmentation in fluorescence images","volume":"95","author":"JC Caicedo","year":"2019","journal-title":"Cytometry A"},{"key":"pcbi.1009879.ref018","doi-asserted-by":"crossref","first-page":"1233","DOI":"10.1038\/s41592-019-0403-1","article-title":"Deep learning for cellular image analysis","volume":"16","author":"E Moen","year":"2019","journal-title":"Nat Methods"},{"key":"pcbi.1009879.ref019","doi-asserted-by":"crossref","first-page":"293","DOI":"10.1109\/CVPR.2017.39","article-title":"End-to-End Instance Segmentation with Recurrent Attention","author":"M Ren","year":"2017","journal-title":"2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE"},{"key":"pcbi.1009879.ref020","article-title":"Convolutional Neural Network-Based Instance Segmentation Algorithm to Acquire Quantitative Criteria of Early Mouse Development","author":"Y Tokuoka","year":"2018","journal-title":"bioRxiv"},{"key":"pcbi.1009879.ref021","first-page":"223","article-title":"IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019)","volume":"2019","author":"D Eschweiler","year":"2019","journal-title":"IEEE"},{"key":"pcbi.1009879.ref022","doi-asserted-by":"crossref","first-page":"67","DOI":"10.1038\/s41592-018-0261-2","article-title":"U-Net: deep learning for cell counting, detection, and morphometry","volume":"16","author":"T Falk","year":"2019","journal-title":"Nat Methods"},{"key":"pcbi.1009879.ref023","article-title":"Segmentation with Residual Attention U-Net and an Edge-Enhancement Approach Preserves Cell Shape Features","author":"N Zhu","year":"2020","journal-title":"arXiv"},{"key":"pcbi.1009879.ref024","doi-asserted-by":"crossref","first-page":"2980","DOI":"10.1109\/ICCV.2017.322","volume-title":"IEEE International Conference on Computer Vision (ICCV)","author":"K He","year":"2017"},{"key":"pcbi.1009879.ref025","doi-asserted-by":"crossref","first-page":"365","DOI":"10.1186\/s12859-018-2375-z","article-title":"A deep learning-based algorithm for 2-D cell segmentation in microscopy images","volume":"19","author":"Y Al-Kofahi","year":"2018","journal-title":"BMC Bioinformatics"},{"key":"pcbi.1009879.ref026","doi-asserted-by":"crossref","DOI":"10.7554\/eLife.57613","article-title":"Accurate and versatile 3D segmentation of plant tissues at cellular resolution","volume":"9","author":"A Wolny","year":"2020","journal-title":"Elife"},{"key":"pcbi.1009879.ref027","article-title":"A deep learning pipeline for nucleus segmentation","author":"G Zaki","year":"2020","journal-title":"bioRxiv"},{"key":"pcbi.1009879.ref028","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1155\/2020\/8351725","article-title":"An Improved Mask R-CNN Model for Multiorgan Segmentation","volume":"2020","author":"J-H Shu","year":"2020","journal-title":"Math Probl Eng"},{"key":"pcbi.1009879.ref029","article-title":"Adapting Mask-RCNN for Automatic Nucleus Segmentation","volume":"00500","author":"J. Johnson","year":"2018","journal-title":"arXiv"},{"key":"pcbi.1009879.ref030","doi-asserted-by":"crossref","first-page":"1141","DOI":"10.1038\/nmeth.4473","article-title":"An objective comparison of cell-tracking algorithms","volume":"14","author":"V Ulman","year":"2017","journal-title":"Nat Methods"},{"key":"pcbi.1009879.ref031","first-page":"450","volume-title":"Computer Vision\u2013ECCV 2018 Workshops","author":"L. Leal-Taix\u00e9","year":"2019","edition":"2"},{"key":"pcbi.1009879.ref032","doi-asserted-by":"crossref","first-page":"100","DOI":"10.1038\/s41592-020-01018-x","article-title":"Cellpose: a generalist algorithm for cellular segmentation","volume":"18","author":"C Stringer","year":"2021","journal-title":"Nat Methods"},{"key":"pcbi.1009879.ref033","article-title":"A Generalized Framework for Agglomerative Clustering of Signed Graphs applied to Instance Segmentation","author":"A Bailoni","year":"2019","journal-title":"arXiv"},{"key":"pcbi.1009879.ref034","doi-asserted-by":"crossref","first-page":"e1008193","DOI":"10.1371\/journal.pcbi.1008193","article-title":"NuSeT: A deep learning tool for reliably separating and analyzing crowded cells","volume":"16","author":"L Yang","year":"2020","journal-title":"PLoS Comput Biol"},{"key":"pcbi.1009879.ref035","doi-asserted-by":"crossref","first-page":"547","DOI":"10.1038\/nmeth.1472","article-title":"Imaging plant growth in 4D: robust tissue reconstruction and lineaging at cell resolution","volume":"7","author":"R Fernandez","year":"2010","journal-title":"Nat Methods"},{"key":"pcbi.1009879.ref036","doi-asserted-by":"crossref","first-page":"E8238","DOI":"10.1073\/pnas.1616768113","article-title":"Cell size and growth regulation in the Arabidopsis thaliana apical stem cell niche","volume":"113","author":"L Willis","year":"2016","journal-title":"Proc Natl Acad Sci U S A"},{"key":"pcbi.1009879.ref037","doi-asserted-by":"crossref","first-page":"20640","DOI":"10.1038\/srep20640","article-title":"Fluorescence microscopy image noise reduction using a stochastically-connected random field model","volume":"6","author":"SA Haider","year":"2016","journal-title":"Sci Rep"},{"key":"pcbi.1009879.ref038","doi-asserted-by":"crossref","first-page":"915","DOI":"10.1109\/ICASSP.2013.6637782","article-title":"Blind restoration of confocal microscopy images in presence of a depth-variant blur and Poisson noise","author":"SB Hadj","year":"2013","journal-title":"2013 IEEE International Conference on Acoustics, Speech and Signal Processing"},{"key":"pcbi.1009879.ref039","article-title":"Fast multi-directional DSLM for confocal detection without striping artifacts","author":"P Ricci","year":"2020","journal-title":"bioRxiv"},{"key":"pcbi.1009879.ref040","author":"P Cheng","year":"1990","journal-title":"Image Contrast in Confocal Light Microscopy"},{"key":"pcbi.1009879.ref041","doi-asserted-by":"crossref","DOI":"10.1126\/science.aar5663","article-title":"Contact area-dependent cell communication and the morphological invariance of ascidian embryogenesis","volume":"369","author":"L Guignard","year":"2020","journal-title":"Science"},{"key":"pcbi.1009879.ref042","doi-asserted-by":"crossref","first-page":"05864","DOI":"10.7554\/eLife.05864","article-title":"MorphoGraphX: A platform for quantifying morphogenesis in 4D","volume":"4","author":"P Barbier de Reuille","year":"2015","journal-title":"Elife"},{"key":"pcbi.1009879.ref043","doi-asserted-by":"crossref","first-page":"2812","DOI":"10.1038\/s41467-019-10668-1","article-title":"MorphoNet: an interactive online morphological browser to explore complex multi-scale data","volume":"10","author":"B Leggio","year":"2019","journal-title":"Nat Commun"},{"key":"pcbi.1009879.ref044","article-title":"A New Marching Cubes Algorithm for Interactive Level Set with Application to MR Image Segmentation","author":"D Feltell","year":"2010","journal-title":"ISVC"},{"key":"pcbi.1009879.ref045","doi-asserted-by":"crossref","first-page":"540","DOI":"10.1016\/j.devcel.2021.01.019","article-title":"A multiscale analysis of early flower development in Arabidopsis provides an integrated view of molecular regulation and growth control","volume":"56","author":"Y Refahi","year":"2021","journal-title":"Dev Cell"},{"key":"pcbi.1009879.ref046","doi-asserted-by":"crossref","first-page":"27","DOI":"10.1109\/83.650848","article-title":"A pyramid approach to subpixel registration based on intensity","volume":"7","author":"P Th\u00e9venaz","year":"1998","journal-title":"IEEE Trans Image Process"},{"key":"pcbi.1009879.ref047","doi-asserted-by":"crossref","first-page":"357","DOI":"10.1038\/s41586-020-2649-2","article-title":"Array programming with NumPy","volume":"585","author":"CR Harris","year":"2020","journal-title":"Nature"},{"key":"pcbi.1009879.ref048","doi-asserted-by":"crossref","first-page":"90","DOI":"10.1109\/MCSE.2007.55","article-title":"Matplotlib: A 2D Graphics Environment","volume":"9","author":"JD Hunter","year":"2007","journal-title":"Comput Sci Eng"},{"key":"pcbi.1009879.ref049","doi-asserted-by":"crossref","first-page":"676","DOI":"10.1038\/nmeth.2019","article-title":"Fiji: an open-source platform for biological-image analysis","volume":"9","author":"J Schindelin","year":"2012","journal-title":"Nat Methods"},{"key":"pcbi.1009879.ref050","article-title":"TensorFlow","author":"M Abadi","year":"2015","journal-title":"Large-Scale Machine Learning on Heterogeneous Systems"},{"issue":"29","key":"pcbi.1009879.ref051","article-title":"Metrics for evaluating 3D medical image segmentation: analysis, selection, and tool","volume":"15","author":"AA Taha","year":"2015","journal-title":"BMC Med Imaging"},{"key":"pcbi.1009879.ref052","article-title":"Outils d\u2019analyse d\u2019images et recalage d\u2019individus pour l\u2019\u00e9tude de la morphogen\u00e8se animale et v\u00e9g\u00e9tale","author":"G. Michelin","year":"2016","journal-title":"Doctoral dissertation"},{"key":"pcbi.1009879.ref053","doi-asserted-by":"crossref","DOI":"10.1523\/ENEURO.0124-18.2018","article-title":"High-Fidelity Imaging in Brain-Wide Structural Studies Using Light-Sheet Microscopy","volume":"5","author":"MC M\u00fcllenbroich","year":"2018","journal-title":"eNeuro"},{"key":"pcbi.1009879.ref054","doi-asserted-by":"crossref","first-page":"70","DOI":"10.1038\/s41377-018-0068-z","article-title":"Attenuation artifacts in light sheet fluorescence microscopy corrected by OPTiSPIM","volume":"7","author":"J Mayer","year":"2018","journal-title":"Light Sci Appl"}],"container-title":["PLOS Computational Biology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1009879","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,4,14]],"date-time":"2022-04-14T13:27:37Z","timestamp":1649942857000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1009879"}},"subtitle":[],"editor":[{"given":"Virginie","family":"Uhlmann","sequence":"first","affiliation":[]}],"short-title":[],"issued":{"date-parts":[[2022,4,14]]},"references-count":54,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2022,4,14]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pcbi.1009879","relation":{"has-preprint":[{"id-type":"doi","id":"10.1101\/2021.06.09.447748","asserted-by":"object"}]},"ISSN":["1553-7358"],"issn-type":[{"value":"1553-7358","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,4,14]]}}}