{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T06:04:39Z","timestamp":1789538679661,"version":"build-2803163510"},"reference-count":60,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T00:00:00Z","timestamp":1789516800000},"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. Artif. Intell."],"abstract":"<jats:p>Three-dimensional branching networks exist throughout biological, natural, and engineered systems as pathways through volumetric space. Segmentation is required to reconstruct these networks for analysis, but minor voxel misclassifications can alter connectivity by falsely disconnecting continuous structures or amplifying spurious branches. Addressing this topological vulnerability requires true 3D segmentation models, since 2D slice-by-slice approaches cannot maintain connectivity across x, y, and z axes. Here, we present DeepBranchAI, a 3D nnU-Net model trained on expert-refined mitochondrial FIB-SEM reference labels as a transferable checkpoint for topology-sensitive segmentation of branching networks. Skeletal-muscle mitochondrial reticulum provides a dense and heterogeneous branching-network training target, and FIB-SEM captures this structure with isotropic nanometer-scale detail. Training on this dataset produced a 3D checkpoint with DSC\u202f=\u202f0.942\u202f\u00b1\u202f0.020 across five-fold mitochondrial FIB-SEM validation, providing the basis for subsequent transfer-learning experiments. Transfer learning was evaluated with five-fold cross-validation on paired image\/reference-label datasets from 3D-IRCADb venous segmentation, Plant CT roots, and AeroPath airway CT, using continuity-sensitive measures including clDice and absolute connected-component error. Across the external datasets, DeepBranchAI-pretrained fine-tuning had higher mean clDice than scratch nnU-Net. Relative differences in mean clDice were 11.7% for 3D-IRCADb, 5.5% for Plant CT roots, and 1.9% for AeroPath. Mean absolute connected-component error was 25.0% lower for Plant CT roots and 11.7% lower for AeroPath. In 3D-IRCADb, DeepBranchAI-pretrained fine-tuning also had higher mean Dice and clDice than VesselFM, 0.679 versus 0.464 and 0.629 versus 0.360, respectively. The transfer results suggest that training on structurally heterogeneous skeletal-muscle mitochondrial networks can produce 3D features useful beyond the original FIB-SEM domain. Code, trained weights, external-transfer scripts, and validation tables are provided as open-source resources.<\/jats:p>","DOI":"10.3389\/frai.2026.1862872","type":"journal-article","created":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T05:31:10Z","timestamp":1789536670000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["DeepBranchAI: a transferable 3D segmentation model for branching networks"],"prefix":"10.3389","volume":"9","author":[{"given":"Alexander V.","family":"Maltsev","sequence":"first","affiliation":[{"name":"Intramural Research Program, National Institute on Aging","place":["Baltimore, MD, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lisa M.","family":"Hartnell","sequence":"additional","affiliation":[{"name":"Intramural Research Program, National Institute on Aging","place":["Baltimore, MD, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Luigi","family":"Ferrucci","sequence":"additional","affiliation":[{"name":"Intramural Research Program, National Institute on Aging","place":["Baltimore, MD, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,16]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"710","DOI":"10.1007\/978-3-031-72111-3_67","article-title":"The centerline-cross entropy loss for vessel-like structure segmentation: better topology consistency without sacrificing accuracy. In medical image computing and computer assisted intervention - MICCAI 2024","volume":"15008","author":"Acebes","year":"2024","journal-title":"Lect. Notes Comput. Sci"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"2424","DOI":"10.1093\/bioinformatics\/btx180","article-title":"Trainable Weka segmentation: a machine learning tool for microscopy pixel classification","volume":"33","author":"Arganda-Carreras","year":"2017","journal-title":"Bioinformatics"},{"key":"ref3","doi-asserted-by":"publisher","first-page":"130","DOI":"10.1038\/20144","article-title":"Size and form in efficient transportation networks","volume":"399","author":"Banavar","year":"1999","journal-title":"Nature"},{"key":"ref4","first-page":"601","volume-title":"Advances in Visual Computing. ISVC 2021","author":"Blekos","year":"2021"},{"key":"ref5","doi-asserted-by":"publisher","first-page":"5","DOI":"10.1023\/A:1010933404324","article-title":"Random forests","volume":"45","author":"Breiman","year":"2001","journal-title":"Mach. Learn."},{"key":"ref6","doi-asserted-by":"publisher","first-page":"3","DOI":"10.1109\/TMI.2022.3203309","article-title":"A persistent homology-based topological loss for CNN-based multiclass segmentation of CMR","volume":"42","author":"Byrne","year":"2023","journal-title":"IEEE Trans. Med. Imaging"},{"key":"ref7","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1016\/j.jsb.2019.03.008","article-title":"Semi-automated 3D segmentation of human skeletal muscle using focused ion beam-scanning electron microscopic images","volume":"207","author":"Caffrey","year":"2019","journal-title":"J. Struct. Biol."},{"key":"ref8","doi-asserted-by":"publisher","first-page":"110530","DOI":"10.1016\/j.compbiomed.2025.110530","article-title":"D2-RD-UNet: a dual-stage dual-class framework with connectivity correction for hepatic vessels segmentation","volume":"195","author":"Cavicchioli","year":"2025","journal-title":"Comput. Biol. Med."},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2411.09593","article-title":"SMILE-UHURA challenge - small vessel segmentation at mesoscopic scale from ultra-high resolution 7T magnetic resonance angiograms","author":"Chatterjee","year":"2024","journal-title":"arXiv:2411.09593"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1904.00625","article-title":"Med3D: Transfer Learning for 3D Medical Image Analysis","author":"Chen","year":"2019","journal-title":"arXiv:1904.00625"},{"key":"ref11","doi-asserted-by":"publisher","first-page":"103280","DOI":"10.1016\/j.media.2024.103280","article-title":"TransUNet: rethinking the U-net architecture design for medical image segmentation through the lens of transformers","volume":"97","author":"Chen","year":"2024","journal-title":"Med. Image Anal."},{"key":"ref12","first-page":"424","volume-title":"Medical Image Computing and Computer-Assisted Intervention - MICCAI 2016, Lecture Notes in Computer Science","author":"\u00c7i\u00e7ek","year":"2016"},{"key":"ref13","doi-asserted-by":"publisher","first-page":"8766","DOI":"10.1109\/TPAMI.2020.3013679","article-title":"A topological loss function for deep-learning based image segmentation using persistent homology","volume":"44","author":"Clough","year":"2022","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref14","doi-asserted-by":"publisher","first-page":"103207","DOI":"10.1016\/j.media.2024.103207","article-title":"MONAI label: a framework for AI-assisted interactive labeling of 3D medical images","volume":"95","author":"Diaz-Pinto","year":"2024","journal-title":"Med. Image Anal."},{"key":"ref15","doi-asserted-by":"publisher","first-page":"1669","DOI":"10.1109\/TPAMI.2018.2835450","article-title":"Large scale image segmentation with structured loss based deep learning for connectome reconstruction","volume":"41","author":"Funke","year":"2019","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref16","doi-asserted-by":"publisher","first-page":"617","DOI":"10.1038\/nature14614","article-title":"Mitochondrial reticulum for cellular energy distribution in muscle","volume":"523","author":"Glancy","year":"2015","journal-title":"Nature"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.14189395","article-title":"Training data - plant root segmentation using 3D-Unet","author":"Harwood","year":"2024","journal-title":"Zenodo."},{"key":"ref18","doi-asserted-by":"publisher","first-page":"203","DOI":"10.1038\/s41592-020-01008-z","article-title":"nnU-net: a self-configuring method for deep learning-based biomedical image segmentation","volume":"18","author":"Isensee","year":"2021","journal-title":"Nat. Methods"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.7892040","article-title":"nnU-net model weights for airway segmentation in murine lungs imaged with light sheet microscopy data","author":"Isensee","year":"2023","journal-title":"Zenodo."},{"key":"ref20","doi-asserted-by":"publisher","first-page":"121905","DOI":"10.1016\/j.eswa.2023.121905","article-title":"Ori-net: orientation-guided neural network for automated coronary arteries segmentation","volume":"238","author":"Jiang","year":"2024","journal-title":"Expert Syst. Appl."},{"key":"ref21","doi-asserted-by":"publisher","first-page":"109045","DOI":"10.1016\/j.compbiomed.2024.109045","article-title":"Segmentation of coronary artery based on discriminative frequency learning and coronary-geometric refinement","volume":"181","author":"Jiang","year":"2024","journal-title":"Comput. Biol. Med."},{"key":"ref22","doi-asserted-by":"publisher","first-page":"1251","DOI":"10.1109\/JBHI.2021.3118104","article-title":"3D graph-connectivity constrained network for hepatic vessel segmentation","volume":"26","author":"Li","year":"2022","journal-title":"IEEE J. Biomed. Health Inform."},{"key":"ref23","doi-asserted-by":"publisher","first-page":"60","DOI":"10.1016\/j.media.2017.07.005","article-title":"A survey on deep learning in medical image analysis","volume":"42","author":"Litjens","year":"2017","journal-title":"Med. Image Anal."},{"key":"ref24","doi-asserted-by":"publisher","first-page":"569","DOI":"10.1038\/s41592-023-01815-0","article-title":"ERnet: a tool for the semantic segmentation and quantitative analysis of endoplasmic reticulum topology","volume":"20","author":"Lu","year":"2023","journal-title":"Nat. Methods"},{"key":"ref25","doi-asserted-by":"publisher","first-page":"654","DOI":"10.1038\/s41467-024-44824-z","article-title":"Segment anything in medical images","volume":"15","author":"Ma","year":"2024","journal-title":"Nat. Commun."},{"key":"ref26","author":"Marron","year":""},{"key":"ref27","doi-asserted-by":"publisher","first-page":"17","DOI":"10.1186\/s13007-015-0060-z","article-title":"Direct comparison of MRI and X-ray CT technologies for 3D imaging of root systems in soil: potential and challenges for root trait quantification","volume":"11","author":"Metzner","year":"2015","journal-title":"Plant Methods"},{"key":"ref28","doi-asserted-by":"publisher","first-page":"101874","DOI":"10.1016\/j.media.2020.101874","article-title":"CS2-net: deep learning segmentation of curvilinear structures in medical imaging","volume":"67","author":"Mou","year":"2021","journal-title":"Med. Image Anal."},{"key":"ref29","doi-asserted-by":"publisher","first-page":"8567","DOI":"10.1364\/OE.17.008567","article-title":"Stripe and ring artifact removal with combined wavelet-Fourier filtering","volume":"17","author":"M\u00fcnch","year":"2009","journal-title":"Opt. Express"},{"key":"ref30","doi-asserted-by":"publisher","first-page":"20190833","DOI":"10.1098\/rsif.2019.0833","article-title":"MicroCT analysis of connectivity in porous structures: optimizing data acquisition and analytical methods in the context of tissue engineering","volume":"17","author":"Nair","year":"2020","journal-title":"J. R. Soc. Interface"},{"key":"ref31","doi-asserted-by":"publisher","first-page":"1584","DOI":"10.1161\/ATVBAHA.124.320672","article-title":"3DVascNet: an automated software for segmentation and quantification of mouse vascular networks in 3D","volume":"44","author":"Narotamo","year":"2024","journal-title":"Arterioscler. Thromb. Vasc. Biol."},{"key":"ref32","doi-asserted-by":"publisher","first-page":"1634","DOI":"10.1038\/s41592-022-01663-4","article-title":"Cellpose 2.0: how to train your own model","volume":"19","author":"Pachitariu","year":"2022","journal-title":"Nat. Methods"},{"key":"ref33","doi-asserted-by":"publisher","first-page":"9","DOI":"10.1016\/j.micron.2014.01.009","article-title":"Exploring the third dimension: volume electron microscopy comes of age","volume":"61","author":"Peddie","year":"2014","journal-title":"Micron"},{"key":"ref34","first-page":"234","volume-title":"Medical Image Computing and Computer-Assisted Intervention - MICCAI 2015, Lecture Notes in Computer Science","author":"Ronneberger","year":"2015"},{"key":"ref35","doi-asserted-by":"publisher","first-page":"1217","DOI":"10.1016\/j.media.2014.07.003","article-title":"Comparing algorithms for automated vessel segmentation in computed tomography scans of the lung: the VESSEL12 study","volume":"18","author":"Rudyanto","year":"2014","journal-title":"Med. Image Anal."},{"key":"ref36","doi-asserted-by":"publisher","first-page":"0076","DOI":"10.34133\/plantphenomics.0076","article-title":"3D U-net segmentation improves root system reconstruction from 3D MRI images in automated and manual virtual reality work flows","volume":"5","author":"Selzner","year":"2023","journal-title":"Plant Phenomics"},{"key":"ref37","doi-asserted-by":"publisher","first-page":"46","DOI":"10.1007\/978-3-031-72111-3_5","article-title":"Centerline boundary dice loss for vascular segmentation. In medical image computing and computer assisted intervention - MICCAI 2024","volume":"15008","author":"Shi","year":"2024","journal-title":"Lect. Notes Comput. Sci"},{"key":"ref38","first-page":"16560","volume-title":"in Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2021)","author":"Shit","year":"2021"},{"key":"ref39","volume-title":"3D Image Reconstruction for Comparison of Algorithm Database: a patient-specific Anatomical and Medical Image Database. Technical Report","author":"Soler","year":"2010"},{"key":"ref40","doi-asserted-by":"publisher","first-page":"1362690","DOI":"10.3389\/fphy.2024.1362690","article-title":"A resource-efficient quantum convolutional neural network","volume":"12","author":"Song","year":"2024","journal-title":"Front. Phys."},{"key":"ref41","doi-asserted-by":"publisher","first-page":"e0311416","DOI":"10.1371\/journal.pone.0311416","article-title":"AeroPath: an airway segmentation benchmark dataset with challenging pathology and baseline method","volume":"19","author":"St\u00f8verud","year":"2024","journal-title":"PLoS One"},{"key":"ref42","doi-asserted-by":"publisher","first-page":"102396","DOI":"10.1016\/j.media.2022.102396","article-title":"Generating 3D TOF-MRA volumes and segmentation labels using generative adversarial networks","volume":"78","author":"Subramaniam","year":"2022","journal-title":"Med. Image Anal."},{"key":"ref43","doi-asserted-by":"publisher","first-page":"29","DOI":"10.1186\/s12880-015-0068-x","article-title":"Metrics for evaluating 3D medical image segmentation: analysis, selection, and tool","volume":"15","author":"Taha","year":"2015","journal-title":"BMC Med. Imaging"},{"key":"ref44","doi-asserted-by":"publisher","first-page":"1299","DOI":"10.1109\/TMI.2016.2535302","article-title":"Convolutional neural networks for medical image analysis: full training or fine tuning?","volume":"35","author":"Tajbakhsh","year":"2016","journal-title":"IEEE Trans. Med. Imaging"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1606.02382","article-title":"Deep learning convolutional networks for multiphoton microscopy vasculature segmentation","author":"Teikari","year":"2016","journal-title":"arXiv:1606.02382"},{"key":"ref46","doi-asserted-by":"publisher","first-page":"592352","DOI":"10.3389\/fnins.2020.592352","article-title":"DeepVesselNet: vessel segmentation, Centerline prediction, and bifurcation detection in 3-D angiographic volumes","volume":"14","author":"Tetteh","year":"2020","journal-title":"Front. Neurosci."},{"key":"ref47","doi-asserted-by":"publisher","first-page":"996","DOI":"10.1016\/j.celrep.2019.01.010","article-title":"Quantitative 3D mapping of the human skeletal muscle mitochondrial network","volume":"26","author":"Vincent","year":"2019","journal-title":"Cell Rep."},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2310.15161","article-title":"SAM-Med3D: towards general-purpose segmentation models for volumetric medical images","author":"Wang","year":"2023","journal-title":"arXiv"},{"key":"ref49","doi-asserted-by":"publisher","first-page":"5915","DOI":"10.1038\/s41467-021-26216-9","article-title":"Annotation-efficient deep learning for automatic medical image segmentation","volume":"12","author":"Wang","year":"2021","journal-title":"Nat. Commun."},{"key":"ref50","doi-asserted-by":"publisher","first-page":"107185","DOI":"10.1016\/j.asoc.2021.107185","article-title":"Deep neural networks for improving physical accuracy of 2D and 3D multi-mineral segmentation of rock micro-CT images","volume":"104","author":"Wang","year":"2021","journal-title":"Appl. Soft Comput."},{"key":"ref51","doi-asserted-by":"publisher","first-page":"e230024","DOI":"10.1148\/ryai.230024","article-title":"TotalSegmentator: robust segmentation of 104 anatomic structures in CT images. Radiology","volume":"5","author":"Wasserthal","year":"2023","journal-title":"Artif. Intell."},{"key":"ref52","doi-asserted-by":"publisher","first-page":"66","DOI":"10.1007\/978-3-030-59722-1_7","article-title":"MitoEM dataset: large-scale 3D mitochondria instance Segmentation from EM images","volume":"12265","author":"Wei","year":"2020","journal-title":"In Medical Image Computing and Computer Assisted Intervention - MICCAI"},{"key":"ref53","doi-asserted-by":"publisher","first-page":"122","DOI":"10.1126\/science.276.5309.122","article-title":"A general model for the origin of allometric scaling laws in biology","volume":"276","author":"West","year":"1997","journal-title":"Science"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2411.17386","article-title":"vesselFM: A Foundation Model for Universal 3D Blood Vessel Segmentation","author":"Wittmann","year":"2024","journal-title":"arXiv:2411.17386"},{"key":"ref55","doi-asserted-by":"publisher","first-page":"vbaf021","DOI":"10.1093\/bioadv\/vbaf021","article-title":"Transfer learning improves performance in volumetric electron microscopy organelle segmentation across tissues. Bioinformatics","volume":"5","author":"Xie","year":"2025","journal-title":"Advances"},{"key":"ref56","doi-asserted-by":"publisher","first-page":"e25916","DOI":"10.7554\/eLife.25916","article-title":"Enhanced FIB-SEM systems for large-volume 3D imaging","volume":"6","author":"Xu","year":"2017","journal-title":"eLife"},{"key":"ref57","doi-asserted-by":"publisher","first-page":"102774","DOI":"10.1016\/j.addma.2022.102774","article-title":"Deep learning-based X-ray computed tomography image reconstruction and prediction of compression behavior of 3D printed lattice structures","volume":"54","author":"Yang","year":"2022","journal-title":"Addit. Manuf."},{"key":"ref58","doi-asserted-by":"publisher","first-page":"102957","DOI":"10.1016\/j.media.2023.102957","article-title":"Multi-site, multi-domain airway tree Modeling (ATM\u201922): a public benchmark for pulmonary airway segmentation","volume":"90","author":"Zhang","year":"2023","journal-title":"Med. Image Anal."},{"key":"ref59","doi-asserted-by":"publisher","first-page":"101840","DOI":"10.1016\/j.media.2020.101840","article-title":"Models Genesis","volume":"67","author":"Zhou","year":"2021","journal-title":"Med. Image Anal."},{"key":"ref60","doi-asserted-by":"publisher","first-page":"055304","DOI":"10.1103\/PhysRevE.106.055304","article-title":"Pore-network extraction using discrete Morse theory: preserving the topology of the pore space","volume":"106","author":"Zubov","year":"2022","journal-title":"Phys. Rev. E"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1862872\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T05:31:12Z","timestamp":1789536672000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1862872\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,16]]},"references-count":60,"alternative-id":["10.3389\/frai.2026.1862872"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1862872","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,16]]},"article-number":"1862872"}}