Automatic Segmentation of Mandibular Ramus and Condyles

Detalhes bibliográficos
Autor(a) principal: Le, Celia
Data de Publicação: 2021
Outros Autores: Deleat-Besson, Romain, Prieto, Juan, Brosset, Serge, Dumont, Maxime, Zhang, Winston, Cevidanes, Lucia, Bianchi, Jonas, Ruellas, Antonio, Gomes, Liliane [UNESP], Gurgel, Marcela, Massaro, Camila, Aliaga-Del Castillo, Aron, Yatabe, Marilia, Benavides, Erika, Soki, Fabiana, Al Turkestani, Najla, Evangelista, Karine, Goncalves, Joao [UNESP], Valladares-Neto, Jose, Alves Garcia Silva, Maria, Chaves, Cauby, Costa, Fabio, Garib, Daniela, Oh, Heesoo, Gryak, Jonathan, Styner, Martin, Fillion-Robin, Jean-Christophe, Paniagua, Beatriz, Najarian, Kayvan, Soroushmehr, Reza
Tipo de documento: Artigo de conferência
Idioma: eng
Título da fonte: Repositório Institucional da UNESP
Texto Completo: http://dx.doi.org/10.1109/EMBC46164.2021.9630727
http://hdl.handle.net/11449/223202
Resumo: In order to diagnose TMJ pathologies, we developed and tested a novel algorithm, MandSeg, that combines image processing and machine learning approaches for automatically segmenting the mandibular condyles and ramus. A deep neural network based on the U-Net architecture was trained for this task, using 109 cone-beam computed tomography (CBCT) scans. The ground truth label maps were manually segmented by clinicians. The U-Net takes 2D slices extracted from the 3D volumetric images. All the 3D scans were cropped depending on their size in order to keep only the mandibular region of interest. The same anatomic cropping region was used for every scan in the dataset. The scans were acquired at different centers with different resolutions. Therefore, we resized all scans to 512×512 in the pre-processing step where we also performed contrast adjustment as the original scans had low contrast. After the pre-processing, around 350 slices were extracted from each scan, and used to train the U-Net model. For the cross-validation, the dataset was divided into 10 folds. The training was performed with 60 epochs, a batch size of 8 and a learning rate of 2×10 -5 . The average performance of the models on the test set presented 0.95 ± 0.05 AUC, 0.93 ± 0.06 sensitivity, 0.9998 ± 0.0001 specificity, 0.9996 ± 0.0003 accuracy, and 0.91 ± 0.03 F1 score. This study findings suggest that fast and efficient CBCT image segmentation of the mandibular condyles and ramus from different clinical data sets and centers can be analyzed effectively. Future studies can now extract radiomic and imaging features as potentially relevant objective diagnostic criteria for TMJ pathologies, such as osteoarthritis (OA). The proposed segmentation will allow large datasets to be analyzed more efficiently for disease classification.
id UNSP_881287c32859f70cb2c8c1b712e47d6d
oai_identifier_str oai:repositorio.unesp.br:11449/223202
network_acronym_str UNSP
network_name_str Repositório Institucional da UNESP
repository_id_str 2946
spelling Automatic Segmentation of Mandibular Ramus and CondylesIn order to diagnose TMJ pathologies, we developed and tested a novel algorithm, MandSeg, that combines image processing and machine learning approaches for automatically segmenting the mandibular condyles and ramus. A deep neural network based on the U-Net architecture was trained for this task, using 109 cone-beam computed tomography (CBCT) scans. The ground truth label maps were manually segmented by clinicians. The U-Net takes 2D slices extracted from the 3D volumetric images. All the 3D scans were cropped depending on their size in order to keep only the mandibular region of interest. The same anatomic cropping region was used for every scan in the dataset. The scans were acquired at different centers with different resolutions. Therefore, we resized all scans to 512×512 in the pre-processing step where we also performed contrast adjustment as the original scans had low contrast. After the pre-processing, around 350 slices were extracted from each scan, and used to train the U-Net model. For the cross-validation, the dataset was divided into 10 folds. The training was performed with 60 epochs, a batch size of 8 and a learning rate of 2×10 -5 . The average performance of the models on the test set presented 0.95 ± 0.05 AUC, 0.93 ± 0.06 sensitivity, 0.9998 ± 0.0001 specificity, 0.9996 ± 0.0003 accuracy, and 0.91 ± 0.03 F1 score. This study findings suggest that fast and efficient CBCT image segmentation of the mandibular condyles and ramus from different clinical data sets and centers can be analyzed effectively. Future studies can now extract radiomic and imaging features as potentially relevant objective diagnostic criteria for TMJ pathologies, such as osteoarthritis (OA). The proposed segmentation will allow large datasets to be analyzed more efficiently for disease classification.School of Dentistry University of MichiganPsychiatry Department University of North CarolinaDepartment of Orthodontics University of the Pacific Arthur A. Dugoni School of DentistryDepartment of Orthodontics Federal University of Rio de JaneiroDepartment of Orthodontics Sao Paulo State UniversityDepartment of Orthodontics Federal University of CearaDepartment of Orthodontics University of Sao PauloDepartment of Orthodontics Federal University of GoiasDepartment of Computational Medicine and Bioinformatics Univ. of MichiganKitware Inc.Department of Orthodontics Sao Paulo State UniversityUniversity of MichiganUniversity of North CarolinaArthur A. Dugoni School of DentistryFederal University of Rio de JaneiroUniversidade Estadual Paulista (UNESP)Federal University of CearaUniversidade de São Paulo (USP)Federal University of GoiasUniv. of MichiganKitware Inc.Le, CeliaDeleat-Besson, RomainPrieto, JuanBrosset, SergeDumont, MaximeZhang, WinstonCevidanes, LuciaBianchi, JonasRuellas, AntonioGomes, Liliane [UNESP]Gurgel, MarcelaMassaro, CamilaAliaga-Del Castillo, AronYatabe, MariliaBenavides, ErikaSoki, FabianaAl Turkestani, NajlaEvangelista, KarineGoncalves, Joao [UNESP]Valladares-Neto, JoseAlves Garcia Silva, MariaChaves, CaubyCosta, FabioGarib, DanielaOh, HeesooGryak, JonathanStyner, MartinFillion-Robin, Jean-ChristophePaniagua, BeatrizNajarian, KayvanSoroushmehr, Reza2022-04-28T19:49:20Z2022-04-28T19:49:20Z2021-01-01info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/conferenceObject2952-2955http://dx.doi.org/10.1109/EMBC46164.2021.9630727Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS, p. 2952-2955.1557-170Xhttp://hdl.handle.net/11449/22320210.1109/EMBC46164.2021.96307272-s2.0-85122499269Scopusreponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengProceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBSinfo:eu-repo/semantics/openAccess2022-04-28T19:49:21Zoai:repositorio.unesp.br:11449/223202Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462024-08-05T15:20:51.684317Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false
dc.title.none.fl_str_mv Automatic Segmentation of Mandibular Ramus and Condyles
title Automatic Segmentation of Mandibular Ramus and Condyles
spellingShingle Automatic Segmentation of Mandibular Ramus and Condyles
Le, Celia
title_short Automatic Segmentation of Mandibular Ramus and Condyles
title_full Automatic Segmentation of Mandibular Ramus and Condyles
title_fullStr Automatic Segmentation of Mandibular Ramus and Condyles
title_full_unstemmed Automatic Segmentation of Mandibular Ramus and Condyles
title_sort Automatic Segmentation of Mandibular Ramus and Condyles
author Le, Celia
author_facet Le, Celia
Deleat-Besson, Romain
Prieto, Juan
Brosset, Serge
Dumont, Maxime
Zhang, Winston
Cevidanes, Lucia
Bianchi, Jonas
Ruellas, Antonio
Gomes, Liliane [UNESP]
Gurgel, Marcela
Massaro, Camila
Aliaga-Del Castillo, Aron
Yatabe, Marilia
Benavides, Erika
Soki, Fabiana
Al Turkestani, Najla
Evangelista, Karine
Goncalves, Joao [UNESP]
Valladares-Neto, Jose
Alves Garcia Silva, Maria
Chaves, Cauby
Costa, Fabio
Garib, Daniela
Oh, Heesoo
Gryak, Jonathan
Styner, Martin
Fillion-Robin, Jean-Christophe
Paniagua, Beatriz
Najarian, Kayvan
Soroushmehr, Reza
author_role author
author2 Deleat-Besson, Romain
Prieto, Juan
Brosset, Serge
Dumont, Maxime
Zhang, Winston
Cevidanes, Lucia
Bianchi, Jonas
Ruellas, Antonio
Gomes, Liliane [UNESP]
Gurgel, Marcela
Massaro, Camila
Aliaga-Del Castillo, Aron
Yatabe, Marilia
Benavides, Erika
Soki, Fabiana
Al Turkestani, Najla
Evangelista, Karine
Goncalves, Joao [UNESP]
Valladares-Neto, Jose
Alves Garcia Silva, Maria
Chaves, Cauby
Costa, Fabio
Garib, Daniela
Oh, Heesoo
Gryak, Jonathan
Styner, Martin
Fillion-Robin, Jean-Christophe
Paniagua, Beatriz
Najarian, Kayvan
Soroushmehr, Reza
author2_role author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
author
dc.contributor.none.fl_str_mv University of Michigan
University of North Carolina
Arthur A. Dugoni School of Dentistry
Federal University of Rio de Janeiro
Universidade Estadual Paulista (UNESP)
Federal University of Ceara
Universidade de São Paulo (USP)
Federal University of Goias
Univ. of Michigan
Kitware Inc.
dc.contributor.author.fl_str_mv Le, Celia
Deleat-Besson, Romain
Prieto, Juan
Brosset, Serge
Dumont, Maxime
Zhang, Winston
Cevidanes, Lucia
Bianchi, Jonas
Ruellas, Antonio
Gomes, Liliane [UNESP]
Gurgel, Marcela
Massaro, Camila
Aliaga-Del Castillo, Aron
Yatabe, Marilia
Benavides, Erika
Soki, Fabiana
Al Turkestani, Najla
Evangelista, Karine
Goncalves, Joao [UNESP]
Valladares-Neto, Jose
Alves Garcia Silva, Maria
Chaves, Cauby
Costa, Fabio
Garib, Daniela
Oh, Heesoo
Gryak, Jonathan
Styner, Martin
Fillion-Robin, Jean-Christophe
Paniagua, Beatriz
Najarian, Kayvan
Soroushmehr, Reza
description In order to diagnose TMJ pathologies, we developed and tested a novel algorithm, MandSeg, that combines image processing and machine learning approaches for automatically segmenting the mandibular condyles and ramus. A deep neural network based on the U-Net architecture was trained for this task, using 109 cone-beam computed tomography (CBCT) scans. The ground truth label maps were manually segmented by clinicians. The U-Net takes 2D slices extracted from the 3D volumetric images. All the 3D scans were cropped depending on their size in order to keep only the mandibular region of interest. The same anatomic cropping region was used for every scan in the dataset. The scans were acquired at different centers with different resolutions. Therefore, we resized all scans to 512×512 in the pre-processing step where we also performed contrast adjustment as the original scans had low contrast. After the pre-processing, around 350 slices were extracted from each scan, and used to train the U-Net model. For the cross-validation, the dataset was divided into 10 folds. The training was performed with 60 epochs, a batch size of 8 and a learning rate of 2×10 -5 . The average performance of the models on the test set presented 0.95 ± 0.05 AUC, 0.93 ± 0.06 sensitivity, 0.9998 ± 0.0001 specificity, 0.9996 ± 0.0003 accuracy, and 0.91 ± 0.03 F1 score. This study findings suggest that fast and efficient CBCT image segmentation of the mandibular condyles and ramus from different clinical data sets and centers can be analyzed effectively. Future studies can now extract radiomic and imaging features as potentially relevant objective diagnostic criteria for TMJ pathologies, such as osteoarthritis (OA). The proposed segmentation will allow large datasets to be analyzed more efficiently for disease classification.
publishDate 2021
dc.date.none.fl_str_mv 2021-01-01
2022-04-28T19:49:20Z
2022-04-28T19:49:20Z
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.driver.fl_str_mv info:eu-repo/semantics/conferenceObject
format conferenceObject
status_str publishedVersion
dc.identifier.uri.fl_str_mv http://dx.doi.org/10.1109/EMBC46164.2021.9630727
Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS, p. 2952-2955.
1557-170X
http://hdl.handle.net/11449/223202
10.1109/EMBC46164.2021.9630727
2-s2.0-85122499269
url http://dx.doi.org/10.1109/EMBC46164.2021.9630727
http://hdl.handle.net/11449/223202
identifier_str_mv Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS, p. 2952-2955.
1557-170X
10.1109/EMBC46164.2021.9630727
2-s2.0-85122499269
dc.language.iso.fl_str_mv eng
language eng
dc.relation.none.fl_str_mv Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv 2952-2955
dc.source.none.fl_str_mv Scopus
reponame:Repositório Institucional da UNESP
instname:Universidade Estadual Paulista (UNESP)
instacron:UNESP
instname_str Universidade Estadual Paulista (UNESP)
instacron_str UNESP
institution UNESP
reponame_str Repositório Institucional da UNESP
collection Repositório Institucional da UNESP
repository.name.fl_str_mv Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)
repository.mail.fl_str_mv
_version_ 1808128499641221120