Noise-Robust Conformal Prediction for Medical Image Classification

Coby Penso; Jacob Goldberger

doi:10.1007/978-3-031-73290-4_16

Noise-Robust Conformal Prediction for Medical Image Classification

Coby Penso, Jacob Goldberger

Bar-Ilan University - The Alexander Kofkin Faculty of Engineering

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

Abstract

Conformal Prediction (CP) quantifies network uncertainty by building a small prediction set with a pre-defined probability that the correct class is within this set. In this study we tackle the problem of CP calibration based on a validation set with noisy labels. We introduce a conformal score that is robust to label noise. The noise-free conformal score is estimated using the noisy labeled data and the noise level. In the test phase the noise-free score is used to form the prediction set. We applied the proposed algorithm to several standard medical imaging classification datasets. We show that our method outperforms current methods by a large margin, in terms of the average size of the prediction set, while maintaining the required coverage.

Original language	English
Title of host publication	Machine Learning in Medical Imaging - 15th International Workshop, MLMI 2024, Held in Conjunction with MICCAI 2024, Proceedings
Editors	Xuanang Xu, Zhiming Cui, Kaicong Sun, Islem Rekik, Xi Ouyang
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	159-168
Number of pages	10
ISBN (Print)	9783031732928
DOIs	https://doi.org/10.1007/978-3-031-73290-4_16
State	Published - 2025
Event	15th International Workshop on Machine Learning in Medical Imaging, MLMI 2024 was held in conjunction with the 27th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2024 - Marrakesh, Morocco Duration: 6 Oct 2024 → 6 Oct 2024

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	15242 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	15th International Workshop on Machine Learning in Medical Imaging, MLMI 2024 was held in conjunction with the 27th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2024
Country/Territory	Morocco
City	Marrakesh
Period	6/10/24 → 6/10/24

Bibliographical note

Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Keywords

conformal prediction
conformal score
label noise
prediction set

Access to Document

10.1007/978-3-031-73290-4_16

Cite this

Penso, C., & Goldberger, J. (2025). Noise-Robust Conformal Prediction for Medical Image Classification. In X. Xu, Z. Cui, K. Sun, I. Rekik, & X. Ouyang (Eds.), Machine Learning in Medical Imaging - 15th International Workshop, MLMI 2024, Held in Conjunction with MICCAI 2024, Proceedings (pp. 159-168). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 15242 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-73290-4_16

Penso, Coby ; Goldberger, Jacob. / Noise-Robust Conformal Prediction for Medical Image Classification. Machine Learning in Medical Imaging - 15th International Workshop, MLMI 2024, Held in Conjunction with MICCAI 2024, Proceedings. editor / Xuanang Xu ; Zhiming Cui ; Kaicong Sun ; Islem Rekik ; Xi Ouyang. Springer Science and Business Media Deutschland GmbH, 2025. pp. 159-168 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{41fa10afab5d4c7694bd2a09d3a96195,

title = "Noise-Robust Conformal Prediction for Medical Image Classification",

abstract = "Conformal Prediction (CP) quantifies network uncertainty by building a small prediction set with a pre-defined probability that the correct class is within this set. In this study we tackle the problem of CP calibration based on a validation set with noisy labels. We introduce a conformal score that is robust to label noise. The noise-free conformal score is estimated using the noisy labeled data and the noise level. In the test phase the noise-free score is used to form the prediction set. We applied the proposed algorithm to several standard medical imaging classification datasets. We show that our method outperforms current methods by a large margin, in terms of the average size of the prediction set, while maintaining the required coverage.",

keywords = "conformal prediction, conformal score, label noise, prediction set",

author = "Coby Penso and Jacob Goldberger",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.; 15th International Workshop on Machine Learning in Medical Imaging, MLMI 2024 was held in conjunction with the 27th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2024 ; Conference date: 06-10-2024 Through 06-10-2024",

year = "2025",

doi = "10.1007/978-3-031-73290-4_16",

language = "אנגלית",

isbn = "9783031732928",

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 = "159--168",

editor = "Xuanang Xu and Zhiming Cui and Kaicong Sun and Islem Rekik and Xi Ouyang",

booktitle = "Machine Learning in Medical Imaging - 15th International Workshop, MLMI 2024, Held in Conjunction with MICCAI 2024, Proceedings",

address = "גרמניה",

}

Penso, C & Goldberger, J 2025, Noise-Robust Conformal Prediction for Medical Image Classification. in X Xu, Z Cui, K Sun, I Rekik & X Ouyang (eds), Machine Learning in Medical Imaging - 15th International Workshop, MLMI 2024, Held in Conjunction with MICCAI 2024, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 15242 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 159-168, 15th International Workshop on Machine Learning in Medical Imaging, MLMI 2024 was held in conjunction with the 27th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2024, Marrakesh, Morocco, 6/10/24. https://doi.org/10.1007/978-3-031-73290-4_16

Noise-Robust Conformal Prediction for Medical Image Classification. / Penso, Coby; Goldberger, Jacob.
Machine Learning in Medical Imaging - 15th International Workshop, MLMI 2024, Held in Conjunction with MICCAI 2024, Proceedings. ed. / Xuanang Xu; Zhiming Cui; Kaicong Sun; Islem Rekik; Xi Ouyang. Springer Science and Business Media Deutschland GmbH, 2025. p. 159-168 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 15242 LNCS).

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

TY - GEN

T1 - Noise-Robust Conformal Prediction for Medical Image Classification

AU - Penso, Coby

AU - Goldberger, Jacob

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

PY - 2025

Y1 - 2025

N2 - Conformal Prediction (CP) quantifies network uncertainty by building a small prediction set with a pre-defined probability that the correct class is within this set. In this study we tackle the problem of CP calibration based on a validation set with noisy labels. We introduce a conformal score that is robust to label noise. The noise-free conformal score is estimated using the noisy labeled data and the noise level. In the test phase the noise-free score is used to form the prediction set. We applied the proposed algorithm to several standard medical imaging classification datasets. We show that our method outperforms current methods by a large margin, in terms of the average size of the prediction set, while maintaining the required coverage.

AB - Conformal Prediction (CP) quantifies network uncertainty by building a small prediction set with a pre-defined probability that the correct class is within this set. In this study we tackle the problem of CP calibration based on a validation set with noisy labels. We introduce a conformal score that is robust to label noise. The noise-free conformal score is estimated using the noisy labeled data and the noise level. In the test phase the noise-free score is used to form the prediction set. We applied the proposed algorithm to several standard medical imaging classification datasets. We show that our method outperforms current methods by a large margin, in terms of the average size of the prediction set, while maintaining the required coverage.

KW - conformal prediction

KW - conformal score

KW - label noise

KW - prediction set

UR - http://www.scopus.com/inward/record.url?scp=85208413490&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-73290-4_16

DO - 10.1007/978-3-031-73290-4_16

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:85208413490

SN - 9783031732928

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

SP - 159

EP - 168

BT - Machine Learning in Medical Imaging - 15th International Workshop, MLMI 2024, Held in Conjunction with MICCAI 2024, Proceedings

A2 - Xu, Xuanang

A2 - Cui, Zhiming

A2 - Sun, Kaicong

A2 - Rekik, Islem

A2 - Ouyang, Xi

PB - Springer Science and Business Media Deutschland GmbH

T2 - 15th International Workshop on Machine Learning in Medical Imaging, MLMI 2024 was held in conjunction with the 27th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2024

Y2 - 6 October 2024 through 6 October 2024

ER -

Penso C, Goldberger J. Noise-Robust Conformal Prediction for Medical Image Classification. In Xu X, Cui Z, Sun K, Rekik I, Ouyang X, editors, Machine Learning in Medical Imaging - 15th International Workshop, MLMI 2024, Held in Conjunction with MICCAI 2024, Proceedings. Springer Science and Business Media Deutschland GmbH. 2025. p. 159-168. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-73290-4_16