Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection

Camille Ruppli; Pietro Gori; Roberto Ardon; Isabelle Bloch

doi:10.1007/978-3-031-44917-8_9

Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection

Camille Ruppli
, Pietro Gori
, Roberto Ardon
, Isabelle Bloch

Institut Polytechnique de Paris
Incepto Medical
Sorbonne Université

Résultats de recherche: Le chapitre dans un livre, un rapport, une anthologie ou une collection › Contribution à une conférence › Revue par des pairs

Résumé

Early diagnosis of prostate cancer is crucial for efficient treatment. Multi-parametric Magnetic Resonance Images (mp-MRI) are widely used for lesion detection. The Prostate Imaging Reporting and Data System (PI-RADS) has standardized interpretation of prostate MRI by defining a score for lesion malignancy. PI-RADS data is readily available from radiology reports but is subject to high inter-reports variability. We propose a new contrastive loss function that leverages weak metadata with multiple annotators per sample and takes advantage of inter-reports variability by defining metadata confidence. By combining metadata of varying confidence with unannotated data into a single conditional contrastive loss function, we report a 3% AUC increase on lesion detection on the public PI-CAI challenge dataset. Code is available at: https://github.com/camilleruppli/decoupled_ccl.

langue originale	Anglais
titre	Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings
rédacteurs en chef	Zhiyun Xue, Sameer Antani, Ghada Zamzmi, Feng Yang, Sivaramakrishnan Rajaraman, Zhaohui Liang, Sharon Xiaolei Huang, Marius George Linguraru
Editeur	Springer Science and Business Media Deutschland GmbH
Pages	95-105
Nombre de pages	11
ISBN (imprimé)	9783031471964
Les DOIs	https://doi.org/10.1007/978-3-031-44917-8_9
état	Publié - 1 janv. 2023
Evénement	2nd Workshop on Medical Image Learning with Noisy and Limited Data, MILLanD 2023 - Vancouver, Canada Durée: 8 oct. 2023 → 8 oct. 2023

Série de publications

Nom	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14307 LNCS
ISSN (imprimé)	0302-9743
ISSN (Electronique)	1611-3349

Une conférence

Une conférence	2nd Workshop on Medical Image Learning with Noisy and Limited Data, MILLanD 2023
Pays/Territoire	Canada
La ville	Vancouver
période	8/10/23 → 8/10/23

SDG des Nations Unies

Ce résultat contribue à ou aux Objectifs de développement durable suivants

SDG 3 Bonne santé et bien-être

Accès au document

10.1007/978-3-031-44917-8_9

Autres fichiers et liens

Lien vers la publication dans Scopus

Contient cette citation

Ruppli, C., Gori, P., Ardon, R., & Bloch, I. (2023). Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection. Dans Z. Xue, S. Antani, G. Zamzmi, F. Yang, S. Rajaraman, Z. Liang, S. X. Huang, & M. G. Linguraru (eds.), Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings (p. 95-105). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol 14307 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-44917-8_9

Ruppli, Camille ; Gori, Pietro ; Ardon, Roberto et al. / Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection. Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings. Editeur / Zhiyun Xue ; Sameer Antani ; Ghada Zamzmi ; Feng Yang ; Sivaramakrishnan Rajaraman ; Zhaohui Liang ; Sharon Xiaolei Huang ; Marius George Linguraru. Springer Science and Business Media Deutschland GmbH, 2023. p. 95-105 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{4c399af8abad4355990b616e02370de7,

title = "Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection",

abstract = "Early diagnosis of prostate cancer is crucial for efficient treatment. Multi-parametric Magnetic Resonance Images (mp-MRI) are widely used for lesion detection. The Prostate Imaging Reporting and Data System (PI-RADS) has standardized interpretation of prostate MRI by defining a score for lesion malignancy. PI-RADS data is readily available from radiology reports but is subject to high inter-reports variability. We propose a new contrastive loss function that leverages weak metadata with multiple annotators per sample and takes advantage of inter-reports variability by defining metadata confidence. By combining metadata of varying confidence with unannotated data into a single conditional contrastive loss function, we report a 3\% AUC increase on lesion detection on the public PI-CAI challenge dataset. Code is available at: https://github.com/camilleruppli/decoupled\_ccl.",

keywords = "Contrastive Learning, Prostate cancer segmentation, Semi-supervised Learning",

author = "Camille Ruppli and Pietro Gori and Roberto Ardon and Isabelle Bloch",

note = "Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 2nd Workshop on Medical Image Learning with Noisy and Limited Data, MILLanD 2023 ; Conference date: 08-10-2023 Through 08-10-2023",

year = "2023",

month = jan,

day = "1",

doi = "10.1007/978-3-031-44917-8\_9",

language = "English",

isbn = "9783031471964",

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 = "95--105",

editor = "Zhiyun Xue and Sameer Antani and Ghada Zamzmi and Feng Yang and Sivaramakrishnan Rajaraman and Zhaohui Liang and Huang, \{Sharon Xiaolei\} and Linguraru, \{Marius George\}",

booktitle = "Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings",

}

Ruppli, C, Gori, P, Ardon, R & Bloch, I 2023, Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection. Dans Z Xue, S Antani, G Zamzmi, F Yang, S Rajaraman, Z Liang, SX Huang & MG Linguraru (eds), Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), VOL. 14307 LNCS, Springer Science and Business Media Deutschland GmbH, p. 95-105, 2nd Workshop on Medical Image Learning with Noisy and Limited Data, MILLanD 2023, Vancouver, Canada, 8/10/23. https://doi.org/10.1007/978-3-031-44917-8_9

Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection. / Ruppli, Camille; Gori, Pietro; Ardon, Roberto et al.
Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings. Ed. / Zhiyun Xue; Sameer Antani; Ghada Zamzmi; Feng Yang; Sivaramakrishnan Rajaraman; Zhaohui Liang; Sharon Xiaolei Huang; Marius George Linguraru. Springer Science and Business Media Deutschland GmbH, 2023. p. 95-105 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol 14307 LNCS).

Résultats de recherche: Le chapitre dans un livre, un rapport, une anthologie ou une collection › Contribution à une conférence › Revue par des pairs

TY - GEN

T1 - Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection

AU - Ruppli, Camille

AU - Gori, Pietro

AU - Ardon, Roberto

AU - Bloch, Isabelle

PY - 2023/1/1

Y1 - 2023/1/1

N2 - Early diagnosis of prostate cancer is crucial for efficient treatment. Multi-parametric Magnetic Resonance Images (mp-MRI) are widely used for lesion detection. The Prostate Imaging Reporting and Data System (PI-RADS) has standardized interpretation of prostate MRI by defining a score for lesion malignancy. PI-RADS data is readily available from radiology reports but is subject to high inter-reports variability. We propose a new contrastive loss function that leverages weak metadata with multiple annotators per sample and takes advantage of inter-reports variability by defining metadata confidence. By combining metadata of varying confidence with unannotated data into a single conditional contrastive loss function, we report a 3% AUC increase on lesion detection on the public PI-CAI challenge dataset. Code is available at: https://github.com/camilleruppli/decoupled_ccl.

AB - Early diagnosis of prostate cancer is crucial for efficient treatment. Multi-parametric Magnetic Resonance Images (mp-MRI) are widely used for lesion detection. The Prostate Imaging Reporting and Data System (PI-RADS) has standardized interpretation of prostate MRI by defining a score for lesion malignancy. PI-RADS data is readily available from radiology reports but is subject to high inter-reports variability. We propose a new contrastive loss function that leverages weak metadata with multiple annotators per sample and takes advantage of inter-reports variability by defining metadata confidence. By combining metadata of varying confidence with unannotated data into a single conditional contrastive loss function, we report a 3% AUC increase on lesion detection on the public PI-CAI challenge dataset. Code is available at: https://github.com/camilleruppli/decoupled_ccl.

KW - Contrastive Learning

KW - Prostate cancer segmentation

KW - Semi-supervised Learning

UR - https://www.scopus.com/pages/publications/85174724831

U2 - 10.1007/978-3-031-44917-8_9

DO - 10.1007/978-3-031-44917-8_9

M3 - Conference contribution

AN - SCOPUS:85174724831

SN - 9783031471964

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

SP - 95

EP - 105

BT - Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings

A2 - Xue, Zhiyun

A2 - Antani, Sameer

A2 - Zamzmi, Ghada

A2 - Yang, Feng

A2 - Rajaraman, Sivaramakrishnan

A2 - Liang, Zhaohui

A2 - Huang, Sharon Xiaolei

A2 - Linguraru, Marius George

PB - Springer Science and Business Media Deutschland GmbH

T2 - 2nd Workshop on Medical Image Learning with Noisy and Limited Data, MILLanD 2023

Y2 - 8 October 2023 through 8 October 2023

ER -

Ruppli C, Gori P, Ardon R, Bloch I. Decoupled Conditional Contrastive Learning with Variable Metadata for Prostate Lesion Detection. Dans Xue Z, Antani S, Zamzmi G, Yang F, Rajaraman S, Liang Z, Huang SX, Linguraru MG, rédacteurs en chef, Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings. Springer Science and Business Media Deutschland GmbH. 2023. p. 95-105. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-44917-8_9