HNT-AI: An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images

Zohaib Salahuddin; Yi Chen; Xian Zhong; Nastaran Mohammadian Rad; Henry C. Woodruff; Philippe Lambin

doi:10.1007/978-3-031-27420-6_21

HNT-AI: An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images

Zohaib Salahuddin^*, Yi Chen, Xian Zhong, Nastaran Mohammadian Rad, Henry C. Woodruff, Philippe Lambin

^*Corresponding author for this work

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

33 Downloads (Pure)

Abstract

Head and neck cancer is one of the most prevalent cancers in the world. Automatic delineation of primary tumors and lymph nodes is important for cancer diagnosis and treatment. In this paper, we develop a deep learning-based model for automatic tumor segmentation, HNT-AI, using PET/CT images provided by the MICCAI 2022 Head and Neck Tumor (HECKTOR) segmentation Challenge. We investigate the effect of residual blocks, squeeze-and-excitation normalization, and grid-attention gates on the performance of 3D-UNET. We project the predicted masks on the z-axis and apply k-means clustering to reduce the number of false positive predictions. Our proposed HNT-AI segmentation framework achieves an aggregated dice score of 0.774 and 0.759 for primary tumors and lymph nodes, respectively, on the unseen external test set. Qualitative analysis of the predicted segmentation masks shows that the predicted segmentation mask tends to follow the high standardized uptake value (SUV) area on the PET scans more closely than the ground truth masks. The largest tumor volume, the larget lymph node volume, and the total number of lymph nodes derived from the segmentation proved to be potential biomarkers for recurrence-free survival with a C-index of 0.627 on the test set.

Original language	English
Title of host publication	Head and Neck Tumor Segmentation and Outcome Prediction
Subtitle of host publication	3rd Challenge, HECKTOR 2022, Held in Conjunction with MICCAI 2022, Proceedings
Editors	Vincent Andrearczyk, Valentin Oreiller, Adrien Depeursinge, Mathieu Hatt
Publisher	Springer Verlag
Pages	212-220
Number of pages	9
Volume	13626 LNCS
Edition	3
ISBN (Electronic)	9783031274206
ISBN (Print)	9783031274190
DOIs	https://doi.org/10.1007/978-3-031-27420-6_21
Publication status	Published - 1 Jan 2023
Event	3rd 3D Head and Neck Tumor Segmentation in PET/CT Challenge, HECKTOR 2022, held in Conjunction with the 25th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2022 - Singapore, Singapore Duration: 22 Sept 2022 → 22 Sept 2022 Conference number: 3

Publication series

Series	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13626 LNCS
ISSN	0302-9743

Conference

Conference	3rd 3D Head and Neck Tumor Segmentation in PET/CT Challenge, HECKTOR 2022, held in Conjunction with the 25th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2022
Country/Territory	Singapore
City	Singapore
Period	22/09/22 → 22/09/22

Keywords

3D UNet
Grid-attention
Residual networks
Segmentation biomarkers
Squeeze-and-excitation

Access to Document

10.1007/978-3-031-27420-6_21

Full TextFinal published version, 617 KBLicence: Taverne

Cite this

Salahuddin, Z., Chen, Y., Zhong, X., Rad, N. M., Woodruff, H. C., & Lambin, P. (2023). HNT-AI: An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images. In V. Andrearczyk, V. Oreiller, A. Depeursinge, & M. Hatt (Eds.), Head and Neck Tumor Segmentation and Outcome Prediction: 3rd Challenge, HECKTOR 2022, Held in Conjunction with MICCAI 2022, Proceedings (3 ed., Vol. 13626 LNCS, pp. 212-220). Springer Verlag. https://doi.org/10.1007/978-3-031-27420-6_21

Salahuddin, Zohaib ; Chen, Yi ; Zhong, Xian et al. / HNT-AI : An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images. Head and Neck Tumor Segmentation and Outcome Prediction: 3rd Challenge, HECKTOR 2022, Held in Conjunction with MICCAI 2022, Proceedings. editor / Vincent Andrearczyk ; Valentin Oreiller ; Adrien Depeursinge ; Mathieu Hatt. Vol. 13626 LNCS 3. ed. Springer Verlag, 2023. pp. 212-220 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 13626 LNCS).

@inproceedings{b7e5ff9db2894dd0a9f243bc4cb058b4,

title = "HNT-AI: An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images",

abstract = "Head and neck cancer is one of the most prevalent cancers in the world. Automatic delineation of primary tumors and lymph nodes is important for cancer diagnosis and treatment. In this paper, we develop a deep learning-based model for automatic tumor segmentation, HNT-AI, using PET/CT images provided by the MICCAI 2022 Head and Neck Tumor (HECKTOR) segmentation Challenge. We investigate the effect of residual blocks, squeeze-and-excitation normalization, and grid-attention gates on the performance of 3D-UNET. We project the predicted masks on the z-axis and apply k-means clustering to reduce the number of false positive predictions. Our proposed HNT-AI segmentation framework achieves an aggregated dice score of 0.774 and 0.759 for primary tumors and lymph nodes, respectively, on the unseen external test set. Qualitative analysis of the predicted segmentation masks shows that the predicted segmentation mask tends to follow the high standardized uptake value (SUV) area on the PET scans more closely than the ground truth masks. The largest tumor volume, the larget lymph node volume, and the total number of lymph nodes derived from the segmentation proved to be potential biomarkers for recurrence-free survival with a C-index of 0.627 on the test set.",

keywords = "3D UNet, Grid-attention, Residual networks, Segmentation biomarkers, Squeeze-and-excitation",

author = "Zohaib Salahuddin and Yi Chen and Xian Zhong and Rad, {Nastaran Mohammadian} and Woodruff, {Henry C.} and Philippe Lambin",

note = "Funding Information: Acknowledgements. We acknowledge financial support from ERC advanced grant (ERC-ADG-2015 n? 694812 - Hypoximmuno), ERC-2020-PoC: 957565-AUTO. DISTINCT. Authors also acknowledge financial support from the European Union{\textquoteright}s Horizon 2020 Research and Innovation Programme under grant agreement: ImmunoSABR n? 733008, MSCA-ITN-PREDICT n? 766276, CHAIMELEON n? 952172, EuCanImage n? 952103, JTI-IMI2-2020-23-two-stage IMI-OPTIMA n? 101034347, and TRANSCAN Joint Transnational Call 2016 (JTC2016 CLEARLY n? UM 2017-8295). This work was supported by the Dutch Cancer Society (KWF Kankerbestrijding), Project number: 14449/2021-PoC. This study has received funding from the National Natural Science Foundation of China (No. 92059201). This study has received funding from the China Scholarship Council (No. 202006675008). Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 3rd 3D Head and Neck Tumor Segmentation in PET/CT Challenge, HECKTOR 2022, held in Conjunction with the 25th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2022 ; Conference date: 22-09-2022 Through 22-09-2022",

year = "2023",

month = jan,

day = "1",

doi = "10.1007/978-3-031-27420-6_21",

language = "English",

isbn = "9783031274190",

volume = "13626 LNCS",

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

publisher = "Springer Verlag",

pages = "212--220",

editor = "Vincent Andrearczyk and Valentin Oreiller and Adrien Depeursinge and Mathieu Hatt",

booktitle = "Head and Neck Tumor Segmentation and Outcome Prediction",

address = "Germany",

edition = "3",

}

Salahuddin, Z , Chen, Y, Zhong, X, Rad, NM , Woodruff, HC & Lambin, P 2023, HNT-AI: An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images. in V Andrearczyk, V Oreiller, A Depeursinge & M Hatt (eds), Head and Neck Tumor Segmentation and Outcome Prediction: 3rd Challenge, HECKTOR 2022, Held in Conjunction with MICCAI 2022, Proceedings. 3 edn, vol. 13626 LNCS, Springer Verlag, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13626 LNCS, pp. 212-220, 3rd 3D Head and Neck Tumor Segmentation in PET/CT Challenge, HECKTOR 2022, held in Conjunction with the 25th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2022, Singapore, Singapore, 22/09/22. https://doi.org/10.1007/978-3-031-27420-6_21

HNT-AI: An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images. / Salahuddin, Zohaib ; Chen, Yi; Zhong, Xian et al.
Head and Neck Tumor Segmentation and Outcome Prediction: 3rd Challenge, HECKTOR 2022, Held in Conjunction with MICCAI 2022, Proceedings. ed. / Vincent Andrearczyk; Valentin Oreiller; Adrien Depeursinge; Mathieu Hatt. Vol. 13626 LNCS 3. ed. Springer Verlag, 2023. p. 212-220 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 13626 LNCS).

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

TY - GEN

T1 - HNT-AI

T2 - 3rd 3D Head and Neck Tumor Segmentation in PET/CT Challenge, HECKTOR 2022, held in Conjunction with the 25th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2022

AU - Salahuddin, Zohaib

AU - Chen, Yi

AU - Zhong, Xian

AU - Rad, Nastaran Mohammadian

AU - Woodruff, Henry C.

AU - Lambin, Philippe

N1 - Conference code: 3

PY - 2023/1/1

Y1 - 2023/1/1

N2 - Head and neck cancer is one of the most prevalent cancers in the world. Automatic delineation of primary tumors and lymph nodes is important for cancer diagnosis and treatment. In this paper, we develop a deep learning-based model for automatic tumor segmentation, HNT-AI, using PET/CT images provided by the MICCAI 2022 Head and Neck Tumor (HECKTOR) segmentation Challenge. We investigate the effect of residual blocks, squeeze-and-excitation normalization, and grid-attention gates on the performance of 3D-UNET. We project the predicted masks on the z-axis and apply k-means clustering to reduce the number of false positive predictions. Our proposed HNT-AI segmentation framework achieves an aggregated dice score of 0.774 and 0.759 for primary tumors and lymph nodes, respectively, on the unseen external test set. Qualitative analysis of the predicted segmentation masks shows that the predicted segmentation mask tends to follow the high standardized uptake value (SUV) area on the PET scans more closely than the ground truth masks. The largest tumor volume, the larget lymph node volume, and the total number of lymph nodes derived from the segmentation proved to be potential biomarkers for recurrence-free survival with a C-index of 0.627 on the test set.

AB - Head and neck cancer is one of the most prevalent cancers in the world. Automatic delineation of primary tumors and lymph nodes is important for cancer diagnosis and treatment. In this paper, we develop a deep learning-based model for automatic tumor segmentation, HNT-AI, using PET/CT images provided by the MICCAI 2022 Head and Neck Tumor (HECKTOR) segmentation Challenge. We investigate the effect of residual blocks, squeeze-and-excitation normalization, and grid-attention gates on the performance of 3D-UNET. We project the predicted masks on the z-axis and apply k-means clustering to reduce the number of false positive predictions. Our proposed HNT-AI segmentation framework achieves an aggregated dice score of 0.774 and 0.759 for primary tumors and lymph nodes, respectively, on the unseen external test set. Qualitative analysis of the predicted segmentation masks shows that the predicted segmentation mask tends to follow the high standardized uptake value (SUV) area on the PET scans more closely than the ground truth masks. The largest tumor volume, the larget lymph node volume, and the total number of lymph nodes derived from the segmentation proved to be potential biomarkers for recurrence-free survival with a C-index of 0.627 on the test set.

KW - 3D UNet

KW - Grid-attention

KW - Residual networks

KW - Segmentation biomarkers

KW - Squeeze-and-excitation

U2 - 10.1007/978-3-031-27420-6_21

DO - 10.1007/978-3-031-27420-6_21

M3 - Conference article in proceeding

SN - 9783031274190

VL - 13626 LNCS

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

SP - 212

EP - 220

BT - Head and Neck Tumor Segmentation and Outcome Prediction

A2 - Andrearczyk, Vincent

A2 - Oreiller, Valentin

A2 - Depeursinge, Adrien

A2 - Hatt, Mathieu

PB - Springer Verlag

Y2 - 22 September 2022 through 22 September 2022

ER -

Salahuddin Z , Chen Y, Zhong X, Rad NM , Woodruff HC , Lambin P. HNT-AI: An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images. In Andrearczyk V, Oreiller V, Depeursinge A, Hatt M, editors, Head and Neck Tumor Segmentation and Outcome Prediction: 3rd Challenge, HECKTOR 2022, Held in Conjunction with MICCAI 2022, Proceedings. 3 ed. Vol. 13626 LNCS. Springer Verlag. 2023. p. 212-220. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 13626 LNCS). doi: 10.1007/978-3-031-27420-6_21