Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge

Chinmay Rao; Suraj Pai; Ibrahim Hadzic; Ivan Zhovannik; Dennis Bontempi; Andre Dekker; Jonas Teuwen; Alberto Traverso

doi:10.1007/978-3-030-67194-5_8

Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge

Chinmay Rao, Suraj Pai, Ibrahim Hadzic, Ivan Zhovannik, Dennis Bontempi, Andre Dekker, Jonas Teuwen, Alberto Traverso

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceeding › Academic › peer-review

74 Downloads (Pure)

Abstract

Automatic segmentation of tumours and organs at risk can function as a useful support tool in radiotherapy treatment planning as well as for validating radiomics studies on larger cohorts. In this paper, we developed robust automatic segmentation methods for the delineation of gross tumour volumes (GTVs) from planning Computed Tomography (CT) and FDG-Positron Emission Tomography (PET) images of head and neck cancer patients. The data was supplied as part of the MICCAI 2020 HECKTOR challenge. We developed two main volumetric approaches: A) an end-to-end volumetric approach and B) a slice-by-slice prediction approach that integrates 3D context around the slice of interest. We exploited differences in the representations provided by these two approaches by ensembling them, obtaining a Dice score of 66.9% on the held out validation set. On an external and independent test set, a final Dice score of 58.7% was achieved.

Original language	English
Title of host publication	Head and Neck Tumor Segmentation. HECKTOR 2020
Editors	V. Andrearczyk, V. Oreiller, A. Depeursinge
Publisher	Springer, Cham
Pages	65-77
Number of pages	13
ISBN (Electronic)	978-3-030-67194-5
ISBN (Print)	978-3-030-67193-8
DOIs	https://doi.org/10.1007/978-3-030-67194-5_8
Publication status	Published - 13 Jan 2021

Publication series

Series	Lecture Notes in Computer Science
Volume	12603
ISSN	0302-9743

Access to Document

10.1007/978-3-030-67194-5_8

Full TextFinal published version, 1.57 MBLicence: Taverne

Cite this

Rao, C., Pai, S., Hadzic, I., Zhovannik, I., Bontempi, D., Dekker, A., Teuwen, J., & Traverso, A. (2021). Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge. In V. Andrearczyk, V. Oreiller, & A. Depeursinge (Eds.), Head and Neck Tumor Segmentation. HECKTOR 2020 (pp. 65-77). Springer, Cham. https://doi.org/10.1007/978-3-030-67194-5_8

@inproceedings{bee3ca1a281f443398d2ad3927960cc5,

title = "Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge",

abstract = "Automatic segmentation of tumours and organs at risk can function as a useful support tool in radiotherapy treatment planning as well as for validating radiomics studies on larger cohorts. In this paper, we developed robust automatic segmentation methods for the delineation of gross tumour volumes (GTVs) from planning Computed Tomography (CT) and FDG-Positron Emission Tomography (PET) images of head and neck cancer patients. The data was supplied as part of the MICCAI 2020 HECKTOR challenge. We developed two main volumetric approaches: A) an end-to-end volumetric approach and B) a slice-by-slice prediction approach that integrates 3D context around the slice of interest. We exploited differences in the representations provided by these two approaches by ensembling them, obtaining a Dice score of 66.9% on the held out validation set. On an external and independent test set, a final Dice score of 58.7% was achieved.",

author = "Chinmay Rao and Suraj Pai and Ibrahim Hadzic and Ivan Zhovannik and Dennis Bontempi and Andre Dekker and Jonas Teuwen and Alberto Traverso",

year = "2021",

month = jan,

day = "13",

doi = "10.1007/978-3-030-67194-5_8",

language = "English",

isbn = "978-3-030-67193-8",

series = "Lecture Notes in Computer Science",

publisher = "Springer, Cham",

pages = "65--77",

editor = "V. Andrearczyk and V. Oreiller and A. Depeursinge",

booktitle = "Head and Neck Tumor Segmentation. HECKTOR 2020",

address = "Switzerland",

}

Rao, C, Pai, S , Hadzic, I, Zhovannik, I, Bontempi, D , Dekker, A, Teuwen, J & Traverso, A 2021, Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge. in V Andrearczyk, V Oreiller & A Depeursinge (eds), Head and Neck Tumor Segmentation. HECKTOR 2020. Springer, Cham, Lecture Notes in Computer Science, vol. 12603, pp. 65-77. https://doi.org/10.1007/978-3-030-67194-5_8

Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge. / Rao, Chinmay; Pai, Suraj ; Hadzic, Ibrahim et al.
Head and Neck Tumor Segmentation. HECKTOR 2020. ed. / V. Andrearczyk; V. Oreiller; A. Depeursinge. Springer, Cham, 2021. p. 65-77 (Lecture Notes in Computer Science, Vol. 12603).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceeding › Academic › peer-review

TY - GEN

T1 - Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge

AU - Rao, Chinmay

AU - Pai, Suraj

AU - Hadzic, Ibrahim

AU - Zhovannik, Ivan

AU - Bontempi, Dennis

AU - Dekker, Andre

AU - Teuwen, Jonas

AU - Traverso, Alberto

PY - 2021/1/13

Y1 - 2021/1/13

N2 - Automatic segmentation of tumours and organs at risk can function as a useful support tool in radiotherapy treatment planning as well as for validating radiomics studies on larger cohorts. In this paper, we developed robust automatic segmentation methods for the delineation of gross tumour volumes (GTVs) from planning Computed Tomography (CT) and FDG-Positron Emission Tomography (PET) images of head and neck cancer patients. The data was supplied as part of the MICCAI 2020 HECKTOR challenge. We developed two main volumetric approaches: A) an end-to-end volumetric approach and B) a slice-by-slice prediction approach that integrates 3D context around the slice of interest. We exploited differences in the representations provided by these two approaches by ensembling them, obtaining a Dice score of 66.9% on the held out validation set. On an external and independent test set, a final Dice score of 58.7% was achieved.

AB - Automatic segmentation of tumours and organs at risk can function as a useful support tool in radiotherapy treatment planning as well as for validating radiomics studies on larger cohorts. In this paper, we developed robust automatic segmentation methods for the delineation of gross tumour volumes (GTVs) from planning Computed Tomography (CT) and FDG-Positron Emission Tomography (PET) images of head and neck cancer patients. The data was supplied as part of the MICCAI 2020 HECKTOR challenge. We developed two main volumetric approaches: A) an end-to-end volumetric approach and B) a slice-by-slice prediction approach that integrates 3D context around the slice of interest. We exploited differences in the representations provided by these two approaches by ensembling them, obtaining a Dice score of 66.9% on the held out validation set. On an external and independent test set, a final Dice score of 58.7% was achieved.

U2 - 10.1007/978-3-030-67194-5_8

DO - 10.1007/978-3-030-67194-5_8

M3 - Conference article in proceeding

SN - 978-3-030-67193-8

T3 - Lecture Notes in Computer Science

SP - 65

EP - 77

BT - Head and Neck Tumor Segmentation. HECKTOR 2020

A2 - Andrearczyk, V.

A2 - Oreiller, V.

A2 - Depeursinge, A.

PB - Springer, Cham

ER -

Rao C, Pai S , Hadzic I, Zhovannik I, Bontempi D , Dekker A et al. Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge. In Andrearczyk V, Oreiller V, Depeursinge A, editors, Head and Neck Tumor Segmentation. HECKTOR 2020. Springer, Cham. 2021. p. 65-77. (Lecture Notes in Computer Science, Vol. 12603). doi: 10.1007/978-3-030-67194-5_8