Meta-learning for Medical Image Segmentation Uncertainty Quantification

Sabri Can Cetindag; Mert Yergin; Deniz Alis; Ilkay Oksuz

doi:10.1007/978-3-031-09002-8_51

Meta-learning for Medical Image Segmentation Uncertainty Quantification

Sabri Can Cetindag^*
, Mert Yergin
, Deniz Alis
, Ilkay Oksuz

^*Bu çalışma için yazışmadan sorumlu yazar

Bilgisayar Mühendisliği Bölümü

Istanbul Technical University
Hevi AI
Acibadem Mehmet Ali Aydinlar Universitesi
King's College London

Araştırma sonucu: Kitap/Rapor/Konferans Bildirisinde Bölüm › Konferans katkısı › bilirkişi

4 Atıf (Scopus)

Özet

Inter-rater and intra-rater variability is a major challenge in medical image segmentation. Inconsistencies of manual segmentations between different experts can challenge development of deterministic automated medical image analysis tools. QUBIQ 2021 is a challenge to enable the successful development of automated machine learning tools, when there are inconsistencies between the labels of different annotators. In this paper, we propose to use meta-learning for quantifying uncertainty in biomedical image quantification. We first train a segmentation network for each expert separately with extensive data augmentation using the nnUnet framework. Then, a meta learner model based on a conventional U-net architecture is trained using the average of all annotators as ground truth, and output of all models that have been trained for each radiologist as input. We compared our results of meta-learning with ensemble methods for various image segmentation tasks and illustrate improved performance.

Orijinal dil	İngilizce
Ana bilgisayar yayını başlığı	Brainlesion
Ana bilgisayar yayını alt yazısı	Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 7th International Workshop, BrainLes 2021, Held in Conjunction with MICCAI 2021, Revised Selected Papers
Editörler	Alessandro Crimi, Spyridon Bakas
Yayınlayan	Springer Science and Business Media Deutschland GmbH
Sayfalar	578-584
Sayfa sayısı	7
ISBN (Basılı)	9783031090011
DOI'lar	https://doi.org/10.1007/978-3-031-09002-8_51
Yayın durumu	Yayınlandı - 2022
Etkinlik	7th International Brain Lesion Workshop, BrainLes 2021, held in conjunction with the Medical Image Computing and Computer Assisted Intervention, MICCAI 2021 - Virtual, Online Süre: 27 Eyl 2021 → 27 Eyl 2021

Yayın serisi

Adı	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Hacim	12963 LNCS
ISSN (Basılı)	0302-9743
ISSN (Elektronik)	1611-3349

???event.eventtypes.event.conference???

???event.eventtypes.event.conference???	7th International Brain Lesion Workshop, BrainLes 2021, held in conjunction with the Medical Image Computing and Computer Assisted Intervention, MICCAI 2021
Şehir	Virtual, Online
Periyot	27/09/21 → 27/09/21

Bibliyografik not

Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Finansman

Acknowledgments. This paper has been produced benefiting from the 2232 International Fellowship for Outstanding Researchers Program of TUBITAK (Project No: 118C353). However, the entire responsibility of the publication/paper belongs to the owner of the paper. The financial support received from TUBITAK does not mean that the content of the publication is approved in a scientific sense by TUBITAK.

Finansörler	Finansör numarası
Türkiye Bilimsel ve Teknolojik Araştırma Kurumu	118C353

Belgeye Erişim

10.1007/978-3-031-09002-8_51

Diğer Dosyalar ve Linkler

Link to publication in Scopus

Alıntı Yap

Cetindag, S. C., Yergin, M., Alis, D., & Oksuz, I. (2022). Meta-learning for Medical Image Segmentation Uncertainty Quantification. In A. Crimi, & S. Bakas (Eds.), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 7th International Workshop, BrainLes 2021, Held in Conjunction with MICCAI 2021, Revised Selected Papers (pp. 578-584). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Hac. 12963 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-09002-8_51

Cetindag, Sabri Can ; Yergin, Mert ; Alis, Deniz et al. / Meta-learning for Medical Image Segmentation Uncertainty Quantification. Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 7th International Workshop, BrainLes 2021, Held in Conjunction with MICCAI 2021, Revised Selected Papers. editör / Alessandro Crimi ; Spyridon Bakas. Springer Science and Business Media Deutschland GmbH, 2022. pp. 578-584 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{e2764f57d0e7473aac69969cf76bf971,

title = "Meta-learning for Medical Image Segmentation Uncertainty Quantification",

abstract = "Inter-rater and intra-rater variability is a major challenge in medical image segmentation. Inconsistencies of manual segmentations between different experts can challenge development of deterministic automated medical image analysis tools. QUBIQ 2021 is a challenge to enable the successful development of automated machine learning tools, when there are inconsistencies between the labels of different annotators. In this paper, we propose to use meta-learning for quantifying uncertainty in biomedical image quantification. We first train a segmentation network for each expert separately with extensive data augmentation using the nnUnet framework. Then, a meta learner model based on a conventional U-net architecture is trained using the average of all annotators as ground truth, and output of all models that have been trained for each radiologist as input. We compared our results of meta-learning with ensemble methods for various image segmentation tasks and illustrate improved performance.",

keywords = "Medical image segmentation, Meta-learning, Uncertainty quantification",

author = "Cetindag, \{Sabri Can\} and Mert Yergin and Deniz Alis and Ilkay Oksuz",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 7th International Brain Lesion Workshop, BrainLes 2021, held in conjunction with the Medical Image Computing and Computer Assisted Intervention, MICCAI 2021 ; Conference date: 27-09-2021 Through 27-09-2021",

year = "2022",

doi = "10.1007/978-3-031-09002-8\_51",

language = "English",

isbn = "9783031090011",

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

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "578--584",

editor = "Alessandro Crimi and Spyridon Bakas",

booktitle = "Brainlesion",

address = "Germany",

}

Cetindag, SC, Yergin, M, Alis, D & Oksuz, I 2022, Meta-learning for Medical Image Segmentation Uncertainty Quantification. in A Crimi & S Bakas (eds), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 7th International Workshop, BrainLes 2021, Held in Conjunction with MICCAI 2021, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), hac. 12963 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 578-584, 7th International Brain Lesion Workshop, BrainLes 2021, held in conjunction with the Medical Image Computing and Computer Assisted Intervention, MICCAI 2021, Virtual, Online, 27/09/21. https://doi.org/10.1007/978-3-031-09002-8_51

Meta-learning for Medical Image Segmentation Uncertainty Quantification. / Cetindag, Sabri Can; Yergin, Mert; Alis, Deniz et al.
Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 7th International Workshop, BrainLes 2021, Held in Conjunction with MICCAI 2021, Revised Selected Papers. ed. / Alessandro Crimi; Spyridon Bakas. Springer Science and Business Media Deutschland GmbH, 2022. p. 578-584 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Hac. 12963 LNCS).

Araştırma sonucu: Kitap/Rapor/Konferans Bildirisinde Bölüm › Konferans katkısı › bilirkişi

TY - GEN

T1 - Meta-learning for Medical Image Segmentation Uncertainty Quantification

AU - Cetindag, Sabri Can

AU - Yergin, Mert

AU - Alis, Deniz

AU - Oksuz, Ilkay

PY - 2022

Y1 - 2022

N2 - Inter-rater and intra-rater variability is a major challenge in medical image segmentation. Inconsistencies of manual segmentations between different experts can challenge development of deterministic automated medical image analysis tools. QUBIQ 2021 is a challenge to enable the successful development of automated machine learning tools, when there are inconsistencies between the labels of different annotators. In this paper, we propose to use meta-learning for quantifying uncertainty in biomedical image quantification. We first train a segmentation network for each expert separately with extensive data augmentation using the nnUnet framework. Then, a meta learner model based on a conventional U-net architecture is trained using the average of all annotators as ground truth, and output of all models that have been trained for each radiologist as input. We compared our results of meta-learning with ensemble methods for various image segmentation tasks and illustrate improved performance.

AB - Inter-rater and intra-rater variability is a major challenge in medical image segmentation. Inconsistencies of manual segmentations between different experts can challenge development of deterministic automated medical image analysis tools. QUBIQ 2021 is a challenge to enable the successful development of automated machine learning tools, when there are inconsistencies between the labels of different annotators. In this paper, we propose to use meta-learning for quantifying uncertainty in biomedical image quantification. We first train a segmentation network for each expert separately with extensive data augmentation using the nnUnet framework. Then, a meta learner model based on a conventional U-net architecture is trained using the average of all annotators as ground truth, and output of all models that have been trained for each radiologist as input. We compared our results of meta-learning with ensemble methods for various image segmentation tasks and illustrate improved performance.

KW - Medical image segmentation

KW - Meta-learning

KW - Uncertainty quantification

UR - https://www.scopus.com/pages/publications/85135159914

U2 - 10.1007/978-3-031-09002-8_51

DO - 10.1007/978-3-031-09002-8_51

M3 - Conference contribution

AN - SCOPUS:85135159914

SN - 9783031090011

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

SP - 578

EP - 584

BT - Brainlesion

A2 - Crimi, Alessandro

A2 - Bakas, Spyridon

PB - Springer Science and Business Media Deutschland GmbH

T2 - 7th International Brain Lesion Workshop, BrainLes 2021, held in conjunction with the Medical Image Computing and Computer Assisted Intervention, MICCAI 2021

Y2 - 27 September 2021 through 27 September 2021

ER -

Cetindag SC, Yergin M, Alis D, Oksuz I. Meta-learning for Medical Image Segmentation Uncertainty Quantification. In Crimi A, Bakas S, editörler, Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 7th International Workshop, BrainLes 2021, Held in Conjunction with MICCAI 2021, Revised Selected Papers. Springer Science and Business Media Deutschland GmbH. 2022. p. 578-584. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-09002-8_51