{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,24]],"date-time":"2026-02-24T08:46:46Z","timestamp":1771922806338,"version":"3.50.1"},"reference-count":46,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,1,21]],"date-time":"2026-01-21T00:00:00Z","timestamp":1768953600000},"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. Bioinform."],"abstract":"<jats:p>Deep learning (DL) enables automated bone segmentation in micro-CT datasets but can struggle to generalize across developmental stages, anatomical regions, and imaging conditions. We present BP-2D-03, which is a revised 2D Bone-Pores segmentation model. It was fitted to a dataset comprising 20 micro-CT scans spanning five mammalian species and 142,960 image patches. To manage the substantially larger and more varied dataset, we developed a DL software interface with modules for training (\u201cBONe DLFit\u201d), prediction (\u201cBONe DLPred\u201d), and evaluation (\u201cBONe IoU\u201d). These tools resolve prior issues such as slice-level data leakage, high memory usage, and limited multi-GPU support. Model performance was evaluated through three analyses. First, 5-fold cross-validation with three seeds per fold evaluated baseline robustness and stability. The model showed generally high mean Intersection-over-Union (IoU) with minimal variation across seeds, but performance varied more across folds related to differences in scan composition. These findings show that the baseline model is stable overall but that predictivity can decline for atypical scans. Second, 30 benchmarking experiments tested how model architecture, encoder backbone, and patch size influence segmentation IoU and computational efficiency. U-Net and UNet++ architectures with simple convolutional backbones (e.g., ResNet-18) achieved the highest IoU values, approaching 0.97. Third, cross-platform experiments confirmed that results are consistent across hardware configurations, operating systems, and implementations (Avizo 3D and standalone). Together, these analyses demonstrate that the BONe DL software delivers robust baseline performance and reproducible results across platforms.<\/jats:p>","DOI":"10.3389\/fbinf.2025.1677527","type":"journal-article","created":{"date-parts":[[2026,1,21]],"date-time":"2026-01-21T06:52:47Z","timestamp":1768978367000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Deep learning software and revised 2D model to segment bone in micro-CT scans"],"prefix":"10.3389","volume":"5","author":[{"given":"Andrew H.","family":"Lee","sequence":"first","affiliation":[{"name":"Department of Anatomy, College of Graduate Studies, Midwestern University","place":["Glendale, AZ, United States"]},{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]},{"name":"College of Veterinary Medicine, Midwestern University","place":["Glendale, AZ, United States"]},{"name":"Core Facilities-Glendale, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Ganesh","family":"Talluri","sequence":"additional","affiliation":[{"name":"BASIS Peoria","place":["Peoria, AZ, United States"]}]},{"given":"Manan","family":"Damani","sequence":"additional","affiliation":[{"name":"Core Facilities-Glendale, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Brandon Vera","family":"Covarrubias","sequence":"additional","affiliation":[{"name":"Department of Anatomy, College of Graduate Studies, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Helena","family":"Hanna","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Jeremy","family":"Chavez","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Julian M.","family":"Moore","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Jacob","family":"Baradarian","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Michael","family":"Molgaard","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Beau","family":"Nielson","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Kalah","family":"Walden","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Thomas L.","family":"Broderick","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]},{"name":"Department of Physiology, College of Graduate Studies, Midwestern University","place":["Glendale, AZ, United States"]}]},{"given":"Layla","family":"Al-Nakkash","sequence":"additional","affiliation":[{"name":"Arizona College of Osteopathic Medicine, Midwestern University","place":["Glendale, AZ, United States"]},{"name":"Department of Physiology, College of Graduate Studies, Midwestern University","place":["Glendale, AZ, United States"]}]}],"member":"1965","published-online":{"date-parts":[[2026,1,21]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"96","DOI":"10.1186\/s40537-023-00765-w","article-title":"Large scale performance analysis of distributed deep learning frameworks for convolutional neural networks","volume":"10","author":"Aach","year":"2023","journal-title":"J. Big Data"},{"key":"B2","doi-asserted-by":"publisher","first-page":"346","DOI":"10.1186\/s12859-023-05462-2","article-title":"Automatic segmentation of large-scale CT image datasets for detailed body composition analysis","volume":"24","author":"Ahmad","year":"2023","journal-title":"BMC Bioinform"},{"key":"B3","doi-asserted-by":"publisher","first-page":"53","DOI":"10.1186\/s40537-021-00444-8","article-title":"Review of deep learning: concepts, CNN architectures, challenges, applications, future directions","volume":"8","author":"Alzubaidi","year":"2021","journal-title":"J. Big Data"},{"key":"B4","doi-asserted-by":"publisher","first-page":"181","DOI":"10.3390\/bioengineering10020181","article-title":"Comparing 3D, 2.5D, and 2D approaches to brain image auto-segmentation","volume":"10","author":"Avesta","year":"2023","journal-title":"Bioengineering"},{"key":"B5","doi-asserted-by":"publisher","first-page":"05391","DOI":"10.48550\/arXiv.2312.05391","article-title":"Loss functions in the era of semantic segmentation: a survey and outlook","volume":"2312","author":"Azad","year":"2023","journal-title":"arXiv"},{"key":"B6","first-page":"161","article-title":"Femur and hip joint","volume-title":"Micro-tomographic atlas of the mouse skeleton","author":"Bab","year":""},{"key":"B7","first-page":"123","article-title":"Humerus and shoulder joint","volume-title":"Micro-tomographic atlas of the mouse skeleton","author":"Bab","year":""},{"key":"B8","doi-asserted-by":"publisher","first-page":"e220232","DOI":"10.1148\/ryai.220232","article-title":"A guide to cross-validation for artificial intelligence in medical imaging","volume":"5","author":"Bradshaw","year":"2023","journal-title":"Radiol. Artif. Intell."},{"key":"B9","doi-asserted-by":"publisher","first-page":"833","DOI":"10.1007\/978-3-030-01234-2_49","article-title":"Encoder-decoder with atrous separable convolution for semantic image segmentation","author":"Chen","year":"2018","journal-title":"arXiv"},{"key":"B10","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2102.04306","article-title":"TransUNet: transformers make strong encoders for medical image segmentation","author":"Chen","year":"2021","journal-title":"arXiv"},{"key":"B11","doi-asserted-by":"publisher","first-page":"2202","DOI":"10.48550\/arXiv.2202.02326","article-title":"Towards training reproducible deep learning models","author":"Chen","year":"2022","journal-title":"arXiv"},{"key":"B12","first-page":"1","article-title":"Are 3D better than 2D convolutional neural networks for medical imaging semantic segmentation?","author":"Crespi","year":"2022"},{"key":"B13","first-page":"248","article-title":"ImageNet: a large-scale hierarchical image database","author":"Deng","year":"2009"},{"key":"B14","doi-asserted-by":"publisher","first-page":"4101","DOI":"10.21105\/joss.04101","article-title":"TorchMetrics - measuring reproducibility in PyTorch","volume":"7","author":"Detlefsen","year":"2022","journal-title":"J. Open Source Softw."},{"key":"B15","doi-asserted-by":"publisher","first-page":"1","DOI":"10.3389\/fninf.2015.00012","article-title":"Reproducibility of neuroimaging analyses across operating systems","volume":"9","author":"Glatard","year":"2015","journal-title":"Front. Neuroinform"},{"key":"B16","doi-asserted-by":"publisher","first-page":"34","DOI":"10.1016\/j.future.2021.06.014","article-title":"Do machine learning platforms provide out-of-the-box reproducibility?","volume":"126","author":"Gundersen","year":"2022","journal-title":"FGCS"},{"key":"B17","doi-asserted-by":"publisher","first-page":"357","DOI":"10.1038\/s41586-020-2649-2","article-title":"Array programming with NumPy","volume":"585","author":"Harris","year":"2020","journal-title":"Nature"},{"key":"B18","first-page":"770","article-title":"Deep residual learning for image recognition","author":"He","year":"2016"},{"key":"B19","first-page":"558","article-title":"Bag of tricks for image classification with convolutional neural networks","author":"He","year":"2019"},{"key":"B20","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1093\/bioinformatics\/btad164","article-title":"mlf-core: a framework for deterministic machine learning","volume":"39","author":"Heumos","year":"2023","journal-title":"Bioinform"},{"key":"B21","unstructured":"Segmentation models pytorch\n          \n          \n            \n              Iakubovskii\n              P.\n            \n          \n          \n          2019"},{"key":"B22","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1801.05746","article-title":"TernausNet: U-net with VGG11 encoder pre-trained on ImageNet for image segmentation","author":"Iglovikov","year":"2018","journal-title":"arXiv 1801.05746v1"},{"key":"B23","unstructured":"Extract and merge image patches (EMPatches)\n          \n          \n            \n              Ilyas\n              T.\n            \n          \n          \n          2023"},{"key":"B24","doi-asserted-by":"publisher","first-page":"37","DOI":"10.1111\/j.1469-8137.1912.tb05611.x","article-title":"The distribution of the flora in the alpine zone","volume":"11","author":"Jaccard","year":"1912","journal-title":"New Phytol."},{"key":"B25","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2015"},{"key":"B26","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1002\/ar.25633","article-title":"Segmentation of cortical bone, trabecular bone, and medullary pores from micro-CT images using 2D and 3D deep learning models","volume":"308","author":"Lee","year":"2025","journal-title":"Anat. Rec."},{"key":"B27","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2509.09267","article-title":"Unified start, personalized end: progressive pruning for efficient 3D medical image segmentation","author":"Li","year":"2025","journal-title":"arXiv"},{"key":"B28","doi-asserted-by":"publisher","first-page":"e1011529","DOI":"10.1371\/journal.pcbi.1011529","article-title":"Natural variability in bee brain size and symmetry revealed by micro-CT imaging and deep learning","volume":"19","author":"L\u00f6sel","year":"2023","journal-title":"PLOS Comput. Biol."},{"key":"B29","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1608.03983","article-title":"SGDR: stochastic gradient descent with warm restarts","author":"Loshchilov","year":"2017","journal-title":"arXiv"},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1711.05101","article-title":"Decoupled weight decay regularization","author":"Loshchilov","year":"2019","journal-title":"arXiv"},{"key":"B31","doi-asserted-by":"publisher","first-page":"81","DOI":"10.3390\/s25010081","article-title":"Data-efficient bone segmentation using feature pyramid-based SegFormer","volume":"25","author":"Masuda","year":"2025","journal-title":"Sensors"},{"key":"B32","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1809.05676","article-title":"Deterministic implementations for reproducibility in deep reinforcement learning","author":"Nagarajan","year":"2019","journal-title":"arXiv"},{"key":"B33","doi-asserted-by":"publisher","first-page":"20220665","DOI":"10.1515\/biol-2022-0665","article-title":"Stochastic gradient descent optimisation for convolutional neural network for medical image segmentation","volume":"18","author":"Nagendram","year":"2023","journal-title":"Open Life Sci."},{"key":"B34","doi-asserted-by":"publisher","first-page":"135334","DOI":"10.1016\/j.jclepro.2022.135334","article-title":"Do we need exotic models? Engineering metrics to enable green machine learning from tackling accuracy-energy trade-offs","volume":"382","author":"Naser","year":"2023","journal-title":"J. Clean. Prod."},{"key":"B35","first-page":"8026","article-title":"PyTorch: an imperative style, high-performance deep learning library","author":"Paszke","year":"2019"},{"key":"B36","doi-asserted-by":"publisher","first-page":"174","DOI":"10.3390\/jimaging11060174","article-title":"A study on energy consumption in AI-driven medical image segmentation","volume":"11","author":"Prajwal","year":"2025","journal-title":"J. Imaging"},{"key":"B37","unstructured":"DataParallel\n          \n          \n          2025"},{"key":"B38","doi-asserted-by":"publisher","first-page":"373","DOI":"10.1177\/002215540305100312","article-title":"Temporal analysis of rat growth plates: cessation of growth with age despite presence of a physis","volume":"51","author":"Roach","year":"2003","journal-title":"J. Histochem. Cytochem."},{"key":"B39","first-page":"234","article-title":"U-Net: Convolutional networks for biomedical image segmentation","volume-title":"Medical image computing and computer-assisted intervention \u2013 MICCAI 2015","author":"Ronneberger","year":"2015"},{"key":"B40","doi-asserted-by":"publisher","first-page":"114","DOI":"10.1016\/S1047-3203(03)00002-6","article-title":"Analysis of image registration noise due to rotationally dependent aliasing","volume":"14","author":"Stone","year":"2003","journal-title":"J. Vis. Commun. Image Represent."},{"key":"B41","first-page":"6105","article-title":"EfficientNet: rethinking model scaling for convolutional neural networks","author":"Tan","year":"2019"},{"key":"B42","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2206.00566","article-title":"The fully convolutional transformer for medical image segmentation","author":"Tragakis","year":"2023","journal-title":"arXiv"},{"key":"B43","first-page":"12077","article-title":"SegFormer: simple and efficient design for semantic segmentation with transformers","author":"Xie","year":"2021"},{"key":"B44","doi-asserted-by":"publisher","first-page":"805271","DOI":"10.3389\/feart.2021.805271","article-title":"CT segmentation of dinosaur fossils by deep learning","volume":"9","author":"Yu","year":"2022","journal-title":"Front. Earth Sci."},{"key":"B45","doi-asserted-by":"publisher","first-page":"102088","DOI":"10.1016\/j.compmedimag.2022.102088","article-title":"Bridging 2D and 3D segmentation networks for computation-efficient volumetric medical image segmentation: an empirical study of 2.5D solutions","volume":"99","author":"Zhang","year":"2022","journal-title":"Comput. Med. Imag. Grap"},{"key":"B46","doi-asserted-by":"publisher","first-page":"3","DOI":"10.1007\/978-3-030-00889-5_1","article-title":"UNet++: a nested U-Net architecture for medical image segmentation","volume":"11045","author":"Zhou","year":"2018","journal-title":"Deep Learn Med. Image Anal. Multimodal Learn Clin. Decis. Support"}],"container-title":["Frontiers in Bioinformatics"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fbinf.2025.1677527\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,24]],"date-time":"2026-02-24T07:41:35Z","timestamp":1771918895000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fbinf.2025.1677527\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,1,21]]},"references-count":46,"alternative-id":["10.3389\/fbinf.2025.1677527"],"URL":"https:\/\/doi.org\/10.3389\/fbinf.2025.1677527","relation":{},"ISSN":["2673-7647"],"issn-type":[{"value":"2673-7647","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,1,21]]},"article-number":"1677527"}}