A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse bone

Abstract Recent advancements in deep learning have significantly enhanced the segmentation of high-resolution microcomputed tomography (µCT) bone scans. In this paper, we present the dual-branch attention-based hybrid network (DBAHNet), a deep learning architecture designed for automatically segment...

Full description

Saved in:
Bibliographic Details
Main Authors: Amine Lagzouli, Peter Pivonka, David M. L. Cooper, Vittorio Sansalone, Alice Othmani
Format: Article
Language:English
Published: Nature Portfolio 2025-03-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-92954-1
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850039598251507712
author Amine Lagzouli
Peter Pivonka
David M. L. Cooper
Vittorio Sansalone
Alice Othmani
author_facet Amine Lagzouli
Peter Pivonka
David M. L. Cooper
Vittorio Sansalone
Alice Othmani
author_sort Amine Lagzouli
collection DOAJ
description Abstract Recent advancements in deep learning have significantly enhanced the segmentation of high-resolution microcomputed tomography (µCT) bone scans. In this paper, we present the dual-branch attention-based hybrid network (DBAHNet), a deep learning architecture designed for automatically segmenting the cortical and trabecular compartments in 3D µCT scans of mouse tibiae. DBAHNet’s hierarchical structure combines transformers and convolutional neural networks to capture long-range dependencies and local features for improved contextual representation. We trained DBAHNet on a limited dataset of 3D µCT scans of mouse tibiae and evaluated its performance on a diverse dataset collected from seven different research studies. This evaluation covered variations in resolutions, ages, mouse strains, drug treatments, surgical procedures, and mechanical loading. DBAHNet demonstrated excellent performance, achieving high accuracy, particularly in challenging scenarios with significantly altered bone morphology. The model’s robustness and generalization capabilities were rigorously tested under diverse and unseen conditions, confirming its effectiveness in the automated segmentation of high-resolution µCT mouse tibia scans. Our findings highlight DBAHNet’s potential to provide reliable and accurate 3D µCT mouse tibia segmentation, thereby enhancing and accelerating preclinical bone studies in drug development. The model and code are available at https://github.com/bigfahma/DBAHNet .
format Article
id doaj-art-e346a3ff3a6b4205b11bf9cbf59ce57c
institution DOAJ
issn 2045-2322
language English
publishDate 2025-03-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-e346a3ff3a6b4205b11bf9cbf59ce57c2025-08-20T02:56:19ZengNature PortfolioScientific Reports2045-23222025-03-0115111810.1038/s41598-025-92954-1A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse boneAmine Lagzouli0Peter Pivonka1David M. L. Cooper2Vittorio Sansalone3Alice Othmani4School of Mechanical, Medical, and Process Engineering, Queensland University of TechnologySchool of Mechanical, Medical, and Process Engineering, Queensland University of TechnologyDepartment of Anatomy, Physiology, and Pharmacology, University of SaskatchewanUniv Paris Est Créteil, Univ Gustave Eiffel, CNRS, UMR 8208, MSMELISSI, Université Paris-Est Creteil (UPEC)Abstract Recent advancements in deep learning have significantly enhanced the segmentation of high-resolution microcomputed tomography (µCT) bone scans. In this paper, we present the dual-branch attention-based hybrid network (DBAHNet), a deep learning architecture designed for automatically segmenting the cortical and trabecular compartments in 3D µCT scans of mouse tibiae. DBAHNet’s hierarchical structure combines transformers and convolutional neural networks to capture long-range dependencies and local features for improved contextual representation. We trained DBAHNet on a limited dataset of 3D µCT scans of mouse tibiae and evaluated its performance on a diverse dataset collected from seven different research studies. This evaluation covered variations in resolutions, ages, mouse strains, drug treatments, surgical procedures, and mechanical loading. DBAHNet demonstrated excellent performance, achieving high accuracy, particularly in challenging scenarios with significantly altered bone morphology. The model’s robustness and generalization capabilities were rigorously tested under diverse and unseen conditions, confirming its effectiveness in the automated segmentation of high-resolution µCT mouse tibia scans. Our findings highlight DBAHNet’s potential to provide reliable and accurate 3D µCT mouse tibia segmentation, thereby enhancing and accelerating preclinical bone studies in drug development. The model and code are available at https://github.com/bigfahma/DBAHNet .https://doi.org/10.1038/s41598-025-92954-13D image segmentationDeep learningMicrocomputed tomography µCTHigh-resolutionMouse tibiaRobust model
spellingShingle Amine Lagzouli
Peter Pivonka
David M. L. Cooper
Vittorio Sansalone
Alice Othmani
A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse bone
Scientific Reports
3D image segmentation
Deep learning
Microcomputed tomography µCT
High-resolution
Mouse tibia
Robust model
title A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse bone
title_full A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse bone
title_fullStr A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse bone
title_full_unstemmed A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse bone
title_short A robust deep learning approach for segmenting cortical and trabecular bone from 3D high resolution µCT scans of mouse bone
title_sort robust deep learning approach for segmenting cortical and trabecular bone from 3d high resolution µct scans of mouse bone
topic 3D image segmentation
Deep learning
Microcomputed tomography µCT
High-resolution
Mouse tibia
Robust model
url https://doi.org/10.1038/s41598-025-92954-1
work_keys_str_mv AT aminelagzouli arobustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT peterpivonka arobustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT davidmlcooper arobustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT vittoriosansalone arobustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT aliceothmani arobustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT aminelagzouli robustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT peterpivonka robustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT davidmlcooper robustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT vittoriosansalone robustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone
AT aliceothmani robustdeeplearningapproachforsegmentingcorticalandtrabecularbonefrom3dhighresolutionμctscansofmousebone