An Investigation into the Impact of Deep Learning Model Choice on Sex and Race Bias in Cardiac MR Segmentation

TY - CHAP

T1 - An Investigation into the Impact of Deep Learning Model Choice on Sex and Race Bias in Cardiac MR Segmentation

AU - Lee, Tiarna

AU - Puyol-Antón, Esther

AU - Ruijsink, Bram

AU - Aitcheson, Keana

AU - Shi, Miaojing

AU - King, Andrew p.

N1 - Funding Information: This work was supported by the Engineering & Physical Sciences Research Council Doctoral Training Partnership (EPSRC DTP) grant EP/T517963/1. This research has been conducted using the UK Biobank Resource under Application Number 17806. Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

PY - 2023/10/9

Y1 - 2023/10/9

N2 - In medical imaging, artificial intelligence (AI) is increasingly being used to automate routine tasks. However, these algorithms can exhibit and exacerbate biases which lead to disparate performances between protected groups. We investigate the impact of model choice on how imbalances in subject sex and race in training datasets affect AI-based cine cardiac magnetic resonance image segmentation. We evaluate three convolutional neural network-based models and one vision transformer model. We find significant sex bias in three of the four models and racial bias in all of the models. However, the severity and nature of the bias varies between the models, highlighting the importance of model choice when attempting to train fair AI-based segmentation models for medical imaging tasks.

AB - In medical imaging, artificial intelligence (AI) is increasingly being used to automate routine tasks. However, these algorithms can exhibit and exacerbate biases which lead to disparate performances between protected groups. We investigate the impact of model choice on how imbalances in subject sex and race in training datasets affect AI-based cine cardiac magnetic resonance image segmentation. We evaluate three convolutional neural network-based models and one vision transformer model. We find significant sex bias in three of the four models and racial bias in all of the models. However, the severity and nature of the bias varies between the models, highlighting the importance of model choice when attempting to train fair AI-based segmentation models for medical imaging tasks.

UR - http://www.scopus.com/inward/record.url?scp=85175854366&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-45249-9_21

DO - 10.1007/978-3-031-45249-9_21

M3 - Chapter

SN - 9783031452482

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

SP - 215

EP - 224

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

ER -