Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction

Yuning Du; Yuyang Xue; Rohan Dharmakumar; Sotirios A. Tsaftaris

doi:10.1007/978-3-031-45249-9_10

Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction

Yuning Du^*, Yuyang Xue, Rohan Dharmakumar, Sotirios A. Tsaftaris

^*Corresponding author for this work

School of Engineering

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

Abstract

Deep learning (DL) reconstruction particularly of MRI has led to improvements in image fidelity and reduction of acquisition time. In neuroimaging, DL methods can reconstruct high-quality images from undersampled data. However, it is essential to consider fairness in DL algorithms, particularly in terms of demographic characteristics. This study presents the first fairness analysis in a DL-based brain MRI reconstruction model. The model utilises the U-Net architecture for image reconstruction and explores the presence and sources of unfairness by implementing baseline Empirical Risk Minimisation (ERM) and rebalancing strategies. Model performance is evaluated using image reconstruction metrics. Our findings reveal statistically significant performance biases between the gender and age subgroups. Surprisingly, data imbalance and training discrimination are not the main sources of bias. This analysis provides insights of fairness in DL-based image reconstruction and aims to improve equity in medical AI applications.

Original language	English
Title of host publication	Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings
Editors	Stefan Wesarg, Cristina Oyarzun Laura, Esther Puyol Antón, Andrew P. King, John S.H. Baxter, Marius Erdt, Klaus Drechsler, Moti Freiman, Yufei Chen, Islem Rekik, Roy Eagleson, Aasa Feragen, Veronika Cheplygina, Melani Ganz-Benjaminsen, Enzo Ferrante, Ben Glocker, Daniel Moyer, Eikel Petersen
Publisher	Springer
Pages	102-111
Number of pages	10
ISBN (Print)	9783031452482
DOIs	https://doi.org/10.1007/978-3-031-45249-9_10
Publication status	Published - 9 Oct 2023
Event	12th International Workshop on Clinical Image-Based Procedures, CLIP 2023, 1st MICCAI Workshop on Fairness of AI in Medical Imaging, FAIMI 2023, held in conjunction with MICCAI 2023 and 2nd MICCAI Workshop on the Ethical and Philosophical Issues in Medical Imaging, EPIMI 2023 - Vancouver, Canada Duration: 12 Oct 2023 → 12 Oct 2023

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14242 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	12th International Workshop on Clinical Image-Based Procedures, CLIP 2023, 1st MICCAI Workshop on Fairness of AI in Medical Imaging, FAIMI 2023, held in conjunction with MICCAI 2023 and 2nd MICCAI Workshop on the Ethical and Philosophical Issues in Medical Imaging, EPIMI 2023
Country/Territory	Canada
City	Vancouver
Period	12/10/23 → 12/10/23

Keywords / Materials (for Non-textual outputs)

Algorithm Bias
Fairness
Image Reconstruction
Neuroimaging

Access to Document

10.1007/978-3-031-45249-9_10

https://arxiv.org/abs/2309.14392Licence: Creative Commons: Attribution (CC-BY)

Cite this

Du, Y., Xue, Y., Dharmakumar, R., & Tsaftaris, S. A. (2023). Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction. In S. Wesarg, C. Oyarzun Laura, E. Puyol Antón, A. P. King, J. S. H. Baxter, M. Erdt, K. Drechsler, M. Freiman, Y. Chen, I. Rekik, R. Eagleson, A. Feragen, V. Cheplygina, M. Ganz-Benjaminsen, E. Ferrante, B. Glocker, D. Moyer, & E. Petersen (Eds.), Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings (pp. 102-111). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14242 LNCS). Springer. https://doi.org/10.1007/978-3-031-45249-9_10

Du, Yuning ; Xue, Yuyang ; Dharmakumar, Rohan et al. / Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction. Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings. editor / Stefan Wesarg ; Cristina Oyarzun Laura ; Esther Puyol Antón ; Andrew P. King ; John S.H. Baxter ; Marius Erdt ; Klaus Drechsler ; Moti Freiman ; Yufei Chen ; Islem Rekik ; Roy Eagleson ; Aasa Feragen ; Veronika Cheplygina ; Melani Ganz-Benjaminsen ; Enzo Ferrante ; Ben Glocker ; Daniel Moyer ; Eikel Petersen. Springer, 2023. pp. 102-111 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{28bdeaf7705340dfbb06dd1d42a6ef50,

title = "Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction",

abstract = "Deep learning (DL) reconstruction particularly of MRI has led to improvements in image fidelity and reduction of acquisition time. In neuroimaging, DL methods can reconstruct high-quality images from undersampled data. However, it is essential to consider fairness in DL algorithms, particularly in terms of demographic characteristics. This study presents the first fairness analysis in a DL-based brain MRI reconstruction model. The model utilises the U-Net architecture for image reconstruction and explores the presence and sources of unfairness by implementing baseline Empirical Risk Minimisation (ERM) and rebalancing strategies. Model performance is evaluated using image reconstruction metrics. Our findings reveal statistically significant performance biases between the gender and age subgroups. Surprisingly, data imbalance and training discrimination are not the main sources of bias. This analysis provides insights of fairness in DL-based image reconstruction and aims to improve equity in medical AI applications.",

keywords = "Algorithm Bias, Fairness, Image Reconstruction, Neuroimaging",

author = "Yuning Du and Yuyang Xue and Rohan Dharmakumar and Tsaftaris, {Sotirios A.}",

year = "2023",

month = oct,

day = "9",

doi = "10.1007/978-3-031-45249-9_10",

language = "English",

isbn = "9783031452482",

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

publisher = "Springer",

pages = "102--111",

editor = "Stefan Wesarg and {Oyarzun Laura}, Cristina and {Puyol Ant{\'o}n}, Esther and King, {Andrew P.} and Baxter, {John S.H.} and Marius Erdt and Klaus Drechsler and Moti Freiman and Yufei Chen and Islem Rekik and Roy Eagleson and Aasa Feragen and Veronika Cheplygina and Melani Ganz-Benjaminsen and Enzo Ferrante and Ben Glocker and Daniel Moyer and Eikel Petersen",

booktitle = "Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings",

address = "United Kingdom",

note = "12th International Workshop on Clinical Image-Based Procedures, CLIP 2023, 1st MICCAI Workshop on Fairness of AI in Medical Imaging, FAIMI 2023, held in conjunction with MICCAI 2023 and 2nd MICCAI Workshop on the Ethical and Philosophical Issues in Medical Imaging, EPIMI 2023 ; Conference date: 12-10-2023 Through 12-10-2023",

}

Du, Y, Xue, Y, Dharmakumar, R & Tsaftaris, SA 2023, Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction. in S Wesarg, C Oyarzun Laura, E Puyol Antón, AP King, JSH Baxter, M Erdt, K Drechsler, M Freiman, Y Chen, I Rekik, R Eagleson, A Feragen, V Cheplygina, M Ganz-Benjaminsen, E Ferrante, B Glocker, D Moyer & E Petersen (eds), Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14242 LNCS, Springer, pp. 102-111, 12th International Workshop on Clinical Image-Based Procedures, CLIP 2023, 1st MICCAI Workshop on Fairness of AI in Medical Imaging, FAIMI 2023, held in conjunction with MICCAI 2023 and 2nd MICCAI Workshop on the Ethical and Philosophical Issues in Medical Imaging, EPIMI 2023, Vancouver, Canada, 12/10/23. https://doi.org/10.1007/978-3-031-45249-9_10

Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction. / Du, Yuning; Xue, Yuyang; Dharmakumar, Rohan et al.
Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings. ed. / Stefan Wesarg; Cristina Oyarzun Laura; Esther Puyol Antón; Andrew P. King; John S.H. Baxter; Marius Erdt; Klaus Drechsler; Moti Freiman; Yufei Chen; Islem Rekik; Roy Eagleson; Aasa Feragen; Veronika Cheplygina; Melani Ganz-Benjaminsen; Enzo Ferrante; Ben Glocker; Daniel Moyer; Eikel Petersen. Springer, 2023. p. 102-111 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14242 LNCS).

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

TY - GEN

T1 - Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction

AU - Du, Yuning

AU - Xue, Yuyang

AU - Dharmakumar, Rohan

AU - Tsaftaris, Sotirios A.

PY - 2023/10/9

Y1 - 2023/10/9

N2 - Deep learning (DL) reconstruction particularly of MRI has led to improvements in image fidelity and reduction of acquisition time. In neuroimaging, DL methods can reconstruct high-quality images from undersampled data. However, it is essential to consider fairness in DL algorithms, particularly in terms of demographic characteristics. This study presents the first fairness analysis in a DL-based brain MRI reconstruction model. The model utilises the U-Net architecture for image reconstruction and explores the presence and sources of unfairness by implementing baseline Empirical Risk Minimisation (ERM) and rebalancing strategies. Model performance is evaluated using image reconstruction metrics. Our findings reveal statistically significant performance biases between the gender and age subgroups. Surprisingly, data imbalance and training discrimination are not the main sources of bias. This analysis provides insights of fairness in DL-based image reconstruction and aims to improve equity in medical AI applications.

AB - Deep learning (DL) reconstruction particularly of MRI has led to improvements in image fidelity and reduction of acquisition time. In neuroimaging, DL methods can reconstruct high-quality images from undersampled data. However, it is essential to consider fairness in DL algorithms, particularly in terms of demographic characteristics. This study presents the first fairness analysis in a DL-based brain MRI reconstruction model. The model utilises the U-Net architecture for image reconstruction and explores the presence and sources of unfairness by implementing baseline Empirical Risk Minimisation (ERM) and rebalancing strategies. Model performance is evaluated using image reconstruction metrics. Our findings reveal statistically significant performance biases between the gender and age subgroups. Surprisingly, data imbalance and training discrimination are not the main sources of bias. This analysis provides insights of fairness in DL-based image reconstruction and aims to improve equity in medical AI applications.

KW - Algorithm Bias

KW - Fairness

KW - Image Reconstruction

KW - Neuroimaging

UR - http://www.scopus.com/inward/record.url?scp=85175848583&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-45249-9_10

DO - 10.1007/978-3-031-45249-9_10

M3 - Conference contribution

AN - SCOPUS:85175848583

SN - 9783031452482

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

SP - 102

EP - 111

BT - Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings

A2 - Wesarg, Stefan

A2 - Oyarzun Laura, Cristina

A2 - Puyol Antón, Esther

A2 - King, Andrew P.

A2 - Baxter, John S.H.

A2 - Erdt, Marius

A2 - Drechsler, Klaus

A2 - Freiman, Moti

A2 - Chen, Yufei

A2 - Rekik, Islem

A2 - Eagleson, Roy

A2 - Feragen, Aasa

A2 - Cheplygina, Veronika

A2 - Ganz-Benjaminsen, Melani

A2 - Ferrante, Enzo

A2 - Glocker, Ben

A2 - Moyer, Daniel

A2 - Petersen, Eikel

PB - Springer

T2 - 12th International Workshop on Clinical Image-Based Procedures, CLIP 2023, 1st MICCAI Workshop on Fairness of AI in Medical Imaging, FAIMI 2023, held in conjunction with MICCAI 2023 and 2nd MICCAI Workshop on the Ethical and Philosophical Issues in Medical Imaging, EPIMI 2023

Y2 - 12 October 2023 through 12 October 2023

ER -

Du Y, Xue Y, Dharmakumar R, Tsaftaris SA. Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction. In Wesarg S, Oyarzun Laura C, Puyol Antón E, King AP, Baxter JSH, Erdt M, Drechsler K, Freiman M, Chen Y, Rekik I, Eagleson R, Feragen A, Cheplygina V, Ganz-Benjaminsen M, Ferrante E, Glocker B, Moyer D, Petersen E, editors, Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings. Springer. 2023. p. 102-111. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-45249-9_10

Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction

Abstract

Publication series

Conference

Keywords / Materials (for Non-textual outputs)

Access to Document

Other files and links

Fingerprint

Canon Medical / RAEng Senior Research Fellow in Healthcare AI

From trivial representations to learning concepts in AI by exploiting unique data

Cite this

Unveiling Fairness Biases in Deep Learning-Based Brain MRI Reconstruction

Abstract

Publication series

Conference

Keywords / Materials (for Non-textual outputs)

Access to Document

Other files and links

Fingerprint

Projects

Canon Medical / RAEng Senior Research Fellow in Healthcare AI

From trivial representations to learning concepts in AI by exploiting unique data

Cite this