Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework

Dong Nie; Li Wang; Roger Trullo; Jianfu Li; Peng Yuan; James Xia; Dinggang Shen

doi:10.1007/978-3-319-67389-9_31

Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework

Dong Nie, Li Wang, Roger Trullo, Jianfu Li, Peng Yuan, James Xia, Dinggang Shen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

21 Citations (Scopus)

Abstract

Computed tomography (CT) is commonly used as a diagnostic and treatment planning imaging modality in craniomaxillofacial (CMF) surgery to correct patient’s bony defects. A major disadvantage of CT is that it emits harmful ionizing radiation to patients during the exam. Magnetic resonance imaging (MRI) is considered to be much safer and noninvasive, and often used to study CMF soft tissues (e.g., temporomandibular joint and brain). However, it is extremely difficult to accurately segment CMF bony structures from MRI since both bone and air appear to be black in MRI, along with low signal-to-noise ratio and partial volume effect. To this end, we proposed a 3D deep-learning based cascade framework to solve these issues. Specifically, a 3D fully convolutional network (FCN) architecture is first adopted to coarsely segment the bony structures. As the coarsely segmented bony structures by FCN tend to be thicker, convolutional neural network (CNN) is further utilized for fine-grained segmentation. To enhance the discriminative ability of the CNN, we particularly concatenate the predicted probability maps from FCN and the original MRI, and feed them together into the CNN to provide more context information for segmentation. Experimental results demonstrate a good performance and also the clinical feasibility of our proposed 3D deep-learning based cascade framework.

Original language	English
Title of host publication	Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings
Editors	Yinghuan Shi, Heung-Il Suk, Kenji Suzuki, Qian Wang
Publisher	Springer Verlag
Pages	266-273
Number of pages	8
ISBN (Print)	9783319673882
DOIs	https://doi.org/10.1007/978-3-319-67389-9_31
Publication status	Published - 2017
Event	8th International Workshop on Machine Learning in Medical Imaging, MLMI 2017 held in conjunction with the 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017 - Quebec City, Canada Duration: 2017 Sept 10 → 2017 Sept 10

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10541 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	8th International Workshop on Machine Learning in Medical Imaging, MLMI 2017 held in conjunction with the 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017
Country/Territory	Canada
City	Quebec City
Period	17/9/10 → 17/9/10

Bibliographical note

Publisher Copyright:
© 2017, Springer International Publishing AG.

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-319-67389-9_31

Cite this

Nie, D., Wang, L., Trullo, R., Li, J., Yuan, P., Xia, J., & Shen, D. (2017). Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework. In Y. Shi, H.-I. Suk, K. Suzuki, & Q. Wang (Eds.), Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings (pp. 266-273). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10541 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-67389-9_31

Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework. / Nie, Dong; Wang, Li; Trullo, Roger et al.
Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings. ed. / Yinghuan Shi; Heung-Il Suk; Kenji Suzuki; Qian Wang. Springer Verlag, 2017. p. 266-273 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10541 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Nie, D, Wang, L, Trullo, R, Li, J, Yuan, P, Xia, J & Shen, D 2017, Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework. in Y Shi, H-I Suk, K Suzuki & Q Wang (eds), Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10541 LNCS, Springer Verlag, pp. 266-273, 8th International Workshop on Machine Learning in Medical Imaging, MLMI 2017 held in conjunction with the 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017, Quebec City, Canada, 17/9/10. https://doi.org/10.1007/978-3-319-67389-9_31

Nie D, Wang L, Trullo R, Li J, Yuan P, Xia J et al. Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework. In Shi Y, Suk HI, Suzuki K, Wang Q, editors, Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings. Springer Verlag. 2017. p. 266-273. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-67389-9_31

Nie, Dong ; Wang, Li ; Trullo, Roger et al. / Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework. Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings. editor / Yinghuan Shi ; Heung-Il Suk ; Kenji Suzuki ; Qian Wang. Springer Verlag, 2017. pp. 266-273 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{22e2f02459344a26a225835c2da1b793,

title = "Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework",

abstract = "Computed tomography (CT) is commonly used as a diagnostic and treatment planning imaging modality in craniomaxillofacial (CMF) surgery to correct patient{\textquoteright}s bony defects. A major disadvantage of CT is that it emits harmful ionizing radiation to patients during the exam. Magnetic resonance imaging (MRI) is considered to be much safer and noninvasive, and often used to study CMF soft tissues (e.g., temporomandibular joint and brain). However, it is extremely difficult to accurately segment CMF bony structures from MRI since both bone and air appear to be black in MRI, along with low signal-to-noise ratio and partial volume effect. To this end, we proposed a 3D deep-learning based cascade framework to solve these issues. Specifically, a 3D fully convolutional network (FCN) architecture is first adopted to coarsely segment the bony structures. As the coarsely segmented bony structures by FCN tend to be thicker, convolutional neural network (CNN) is further utilized for fine-grained segmentation. To enhance the discriminative ability of the CNN, we particularly concatenate the predicted probability maps from FCN and the original MRI, and feed them together into the CNN to provide more context information for segmentation. Experimental results demonstrate a good performance and also the clinical feasibility of our proposed 3D deep-learning based cascade framework.",

author = "Dong Nie and Li Wang and Roger Trullo and Jianfu Li and Peng Yuan and James Xia and Dinggang Shen",

note = "Publisher Copyright: {\textcopyright} 2017, Springer International Publishing AG.; 8th International Workshop on Machine Learning in Medical Imaging, MLMI 2017 held in conjunction with the 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017 ; Conference date: 10-09-2017 Through 10-09-2017",

year = "2017",

doi = "10.1007/978-3-319-67389-9_31",

language = "English",

isbn = "9783319673882",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "266--273",

editor = "Yinghuan Shi and Heung-Il Suk and Kenji Suzuki and Qian Wang",

booktitle = "Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings",

}

TY - GEN

T1 - Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework

AU - Nie, Dong

AU - Wang, Li

AU - Trullo, Roger

AU - Li, Jianfu

AU - Yuan, Peng

AU - Xia, James

AU - Shen, Dinggang

PY - 2017

Y1 - 2017

N2 - Computed tomography (CT) is commonly used as a diagnostic and treatment planning imaging modality in craniomaxillofacial (CMF) surgery to correct patient’s bony defects. A major disadvantage of CT is that it emits harmful ionizing radiation to patients during the exam. Magnetic resonance imaging (MRI) is considered to be much safer and noninvasive, and often used to study CMF soft tissues (e.g., temporomandibular joint and brain). However, it is extremely difficult to accurately segment CMF bony structures from MRI since both bone and air appear to be black in MRI, along with low signal-to-noise ratio and partial volume effect. To this end, we proposed a 3D deep-learning based cascade framework to solve these issues. Specifically, a 3D fully convolutional network (FCN) architecture is first adopted to coarsely segment the bony structures. As the coarsely segmented bony structures by FCN tend to be thicker, convolutional neural network (CNN) is further utilized for fine-grained segmentation. To enhance the discriminative ability of the CNN, we particularly concatenate the predicted probability maps from FCN and the original MRI, and feed them together into the CNN to provide more context information for segmentation. Experimental results demonstrate a good performance and also the clinical feasibility of our proposed 3D deep-learning based cascade framework.

AB - Computed tomography (CT) is commonly used as a diagnostic and treatment planning imaging modality in craniomaxillofacial (CMF) surgery to correct patient’s bony defects. A major disadvantage of CT is that it emits harmful ionizing radiation to patients during the exam. Magnetic resonance imaging (MRI) is considered to be much safer and noninvasive, and often used to study CMF soft tissues (e.g., temporomandibular joint and brain). However, it is extremely difficult to accurately segment CMF bony structures from MRI since both bone and air appear to be black in MRI, along with low signal-to-noise ratio and partial volume effect. To this end, we proposed a 3D deep-learning based cascade framework to solve these issues. Specifically, a 3D fully convolutional network (FCN) architecture is first adopted to coarsely segment the bony structures. As the coarsely segmented bony structures by FCN tend to be thicker, convolutional neural network (CNN) is further utilized for fine-grained segmentation. To enhance the discriminative ability of the CNN, we particularly concatenate the predicted probability maps from FCN and the original MRI, and feed them together into the CNN to provide more context information for segmentation. Experimental results demonstrate a good performance and also the clinical feasibility of our proposed 3D deep-learning based cascade framework.

UR - http://www.scopus.com/inward/record.url?scp=85029669526&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-67389-9_31

DO - 10.1007/978-3-319-67389-9_31

M3 - Conference contribution

AN - SCOPUS:85029669526

SN - 9783319673882

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 266

EP - 273

BT - Machine Learning in Medical Imaging - 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Proceedings

A2 - Shi, Yinghuan

A2 - Suk, Heung-Il

A2 - Suzuki, Kenji

A2 - Wang, Qian

PB - Springer Verlag

T2 - 8th International Workshop on Machine Learning in Medical Imaging, MLMI 2017 held in conjunction with the 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017

Y2 - 10 September 2017 through 10 September 2017

ER -

Segmentation of craniomaxillofacial bony structures from MRI with a 3D deep-learning based cascade framework

Abstract

Publication series

Other

Bibliographical note

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this