A Systematic Study of Race and Sex Bias in CNN-Based Cardiac MR Segmentation

TY - CHAP

T1 - A Systematic Study of Race and Sex Bias in CNN-Based Cardiac MR Segmentation

AU - Lee, Tiarna

AU - Puyol-Antón, Esther

AU - Ruijsink, Bram

AU - Shi, Miaojing

AU - King, Andrew p.

N1 - Funding Information: Acknowledgements. 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. 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: © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

PY - 2023/1/28

Y1 - 2023/1/28

N2 - In computer vision there has been significant research interest in assessing potential demographic bias in deep learning models. One of the main causes of such bias is imbalance in the training data. In medical imaging, where the potential impact of bias is arguably much greater, there has been less interest. In medical imaging pipelines, segmentation of structures of interest plays an important role in estimating clinical biomarkers that are subsequently used to inform patient management. Convolutional neural networks (CNNs) are starting to be used to automate this process. We present the first systematic study of the impact of training set imbalance on race and sex bias in CNN-based segmentation. We focus on segmentation of the structures of the heart from short axis cine cardiac magnetic resonance images, and train multiple CNN segmentation models with different levels of race/sex imbalance. We find no significant bias in the sex experiment but significant bias in two separate race experiments, highlighting the need to consider adequate representation of different demographic groups in health datasets.

AB - In computer vision there has been significant research interest in assessing potential demographic bias in deep learning models. One of the main causes of such bias is imbalance in the training data. In medical imaging, where the potential impact of bias is arguably much greater, there has been less interest. In medical imaging pipelines, segmentation of structures of interest plays an important role in estimating clinical biomarkers that are subsequently used to inform patient management. Convolutional neural networks (CNNs) are starting to be used to automate this process. We present the first systematic study of the impact of training set imbalance on race and sex bias in CNN-based segmentation. We focus on segmentation of the structures of the heart from short axis cine cardiac magnetic resonance images, and train multiple CNN segmentation models with different levels of race/sex imbalance. We find no significant bias in the sex experiment but significant bias in two separate race experiments, highlighting the need to consider adequate representation of different demographic groups in health datasets.

UR - http://www.scopus.com/inward/record.url?scp=85148037870&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-23443-9_22

DO - 10.1007/978-3-031-23443-9_22

M3 - Chapter

SN - 9783031234422

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

SP - 233

EP - 244

BT - Statistical Atlases and Computational Models of the Heart. Regular and CMRxMotion Challenge Papers - 13th International Workshop, STACOM 2022, Held in Conjunction with MICCAI 2022, Revised Selected Papers

A2 - Camara, Oscar

A2 - Puyol-Antón, Esther

A2 - Suinesiaputra, Avan

A2 - Young, Alistair

A2 - Qin, Chen

A2 - Sermesant, Maxime

A2 - Wang, Shuo

ER -