Comparison of 2D, 2.5D, and 3D segmentation networks for mandibular canals in CBCT images: a study on public and external datasets

Abstract The purpose of this study was to compare the performances of 2D, 2.5D, and 3D CNN-based segmentation networks, along with a 3D vision transformer-based segmentation network, for segmenting mandibular canals (MCs) on the public and external CBCT datasets under the same GPU memory capacity. W...

Full description

Saved in:
Bibliographic Details
Main Authors: Su Yang, Jong Soo Jeong, Dahyun Song, Ji Yong Han, Sang-Heon Lim, Sujeong Kim, Ji-Yong Yoo, Jun-Min Kim, Jo-Eun Kim, Kyung-Hoe Huh, Sam-Sun Lee, Min-Suk Heo, Won-Jin Yi
Format: Article
Language:English
Published: BMC 2025-07-01
Series:BMC Oral Health
Subjects:
Online Access:https://doi.org/10.1186/s12903-025-06483-4
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849738561815838720
author Su Yang
Jong Soo Jeong
Dahyun Song
Ji Yong Han
Sang-Heon Lim
Sujeong Kim
Ji-Yong Yoo
Jun-Min Kim
Jo-Eun Kim
Kyung-Hoe Huh
Sam-Sun Lee
Min-Suk Heo
Won-Jin Yi
author_facet Su Yang
Jong Soo Jeong
Dahyun Song
Ji Yong Han
Sang-Heon Lim
Sujeong Kim
Ji-Yong Yoo
Jun-Min Kim
Jo-Eun Kim
Kyung-Hoe Huh
Sam-Sun Lee
Min-Suk Heo
Won-Jin Yi
author_sort Su Yang
collection DOAJ
description Abstract The purpose of this study was to compare the performances of 2D, 2.5D, and 3D CNN-based segmentation networks, along with a 3D vision transformer-based segmentation network, for segmenting mandibular canals (MCs) on the public and external CBCT datasets under the same GPU memory capacity. We also performed ablation studies for an image-cropping (IC) technique and segmentation loss functions. 3D-UNet showed the highest segmentation performance for the MC than those of 2D and 2.5D segmentation networks on public test datasets, achieving 0.569 ± 0.107, 0.719 ± 0.092, 0.664 ± 0.131, and 0.812 ± 0.095 in terms of JI, DSC, PR, and RC, respectively. On the external test dataset, 3D-UNet achieved 0.564 ± 0.092, 0.716 ± 0.081, 0.812 ± 0.087, and 0.652 ± 0.103 in terms of JI, DSC, PR, and RC, respectively. The IC technique and multi-planar Dice loss improved the boundary details and structural connectivity of the MC from the mental foramen to the mandibular foramen. The 3D-UNet demonstrated superior segmentation performance for the MC by learning 3D volumetric context information for the entire MC in the CBCT volume.
format Article
id doaj-art-4cb8f62f4f4e43af8d351c868da382fe
institution DOAJ
issn 1472-6831
language English
publishDate 2025-07-01
publisher BMC
record_format Article
series BMC Oral Health
spelling doaj-art-4cb8f62f4f4e43af8d351c868da382fe2025-08-20T03:06:31ZengBMCBMC Oral Health1472-68312025-07-0125111710.1186/s12903-025-06483-4Comparison of 2D, 2.5D, and 3D segmentation networks for mandibular canals in CBCT images: a study on public and external datasetsSu Yang0Jong Soo Jeong1Dahyun Song2Ji Yong Han3Sang-Heon Lim4Sujeong Kim5Ji-Yong Yoo6Jun-Min Kim7Jo-Eun Kim8Kyung-Hoe Huh9Sam-Sun Lee10Min-Suk Heo11Won-Jin Yi12Department of Applied Bioengineering, Graduate School of Convergence Science and Technology, Seoul National UniversityDepartment of Dentistry, School of Dentistry, Seoul National UniversityInterdisciplinary Program in Bioengineering, Graduate School of Engineering, Seoul National UniversityInterdisciplinary Program in Bioengineering, Graduate School of Engineering, Seoul National UniversityInterdisciplinary Program in Bioengineering, Graduate School of Engineering, Seoul National UniversityInterdisciplinary Program in Bioengineering, Graduate School of Engineering, Seoul National UniversityDepartment of Applied Bioengineering, Graduate School of Convergence Science and Technology, Seoul National UniversityDepartment of Electronics and Information Engineering, Hansung UniversityDepartment of Oral and Maxillofacial Radiology and Dental Research Institute, School of Dentistry, Seoul National UniversityDepartment of Oral and Maxillofacial Radiology and Dental Research Institute, School of Dentistry, Seoul National UniversityDepartment of Oral and Maxillofacial Radiology and Dental Research Institute, School of Dentistry, Seoul National UniversityDepartment of Oral and Maxillofacial Radiology and Dental Research Institute, School of Dentistry, Seoul National UniversityDepartment of Applied Bioengineering, Graduate School of Convergence Science and Technology, Seoul National UniversityAbstract The purpose of this study was to compare the performances of 2D, 2.5D, and 3D CNN-based segmentation networks, along with a 3D vision transformer-based segmentation network, for segmenting mandibular canals (MCs) on the public and external CBCT datasets under the same GPU memory capacity. We also performed ablation studies for an image-cropping (IC) technique and segmentation loss functions. 3D-UNet showed the highest segmentation performance for the MC than those of 2D and 2.5D segmentation networks on public test datasets, achieving 0.569 ± 0.107, 0.719 ± 0.092, 0.664 ± 0.131, and 0.812 ± 0.095 in terms of JI, DSC, PR, and RC, respectively. On the external test dataset, 3D-UNet achieved 0.564 ± 0.092, 0.716 ± 0.081, 0.812 ± 0.087, and 0.652 ± 0.103 in terms of JI, DSC, PR, and RC, respectively. The IC technique and multi-planar Dice loss improved the boundary details and structural connectivity of the MC from the mental foramen to the mandibular foramen. The 3D-UNet demonstrated superior segmentation performance for the MC by learning 3D volumetric context information for the entire MC in the CBCT volume.https://doi.org/10.1186/s12903-025-06483-4Deep learningCBCT imageMandibular canalImage segmentation3D segmentation network
spellingShingle Su Yang
Jong Soo Jeong
Dahyun Song
Ji Yong Han
Sang-Heon Lim
Sujeong Kim
Ji-Yong Yoo
Jun-Min Kim
Jo-Eun Kim
Kyung-Hoe Huh
Sam-Sun Lee
Min-Suk Heo
Won-Jin Yi
Comparison of 2D, 2.5D, and 3D segmentation networks for mandibular canals in CBCT images: a study on public and external datasets
BMC Oral Health
Deep learning
CBCT image
Mandibular canal
Image segmentation
3D segmentation network
title Comparison of 2D, 2.5D, and 3D segmentation networks for mandibular canals in CBCT images: a study on public and external datasets
title_full Comparison of 2D, 2.5D, and 3D segmentation networks for mandibular canals in CBCT images: a study on public and external datasets
title_fullStr Comparison of 2D, 2.5D, and 3D segmentation networks for mandibular canals in CBCT images: a study on public and external datasets
title_full_unstemmed Comparison of 2D, 2.5D, and 3D segmentation networks for mandibular canals in CBCT images: a study on public and external datasets
title_short Comparison of 2D, 2.5D, and 3D segmentation networks for mandibular canals in CBCT images: a study on public and external datasets
title_sort comparison of 2d 2 5d and 3d segmentation networks for mandibular canals in cbct images a study on public and external datasets
topic Deep learning
CBCT image
Mandibular canal
Image segmentation
3D segmentation network
url https://doi.org/10.1186/s12903-025-06483-4
work_keys_str_mv AT suyang comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT jongsoojeong comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT dahyunsong comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT jiyonghan comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT sangheonlim comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT sujeongkim comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT jiyongyoo comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT junminkim comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT joeunkim comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT kyunghoehuh comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT samsunlee comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT minsukheo comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets
AT wonjinyi comparisonof2d25dand3dsegmentationnetworksformandibularcanalsincbctimagesastudyonpublicandexternaldatasets