An investigation into the causes of race bias in artificial intelligence–based cine cardiac magnetic resonance segmentation
Distributional differences between annotated CMR data of White and Black races, which can lead to bias in trained AI segmentation models, are predominantly image-based, not segmentation-based. Most of the differences occur in areas outside the heart, such as subcutaneous fat. These findings will be important for researchers investigating performance of AI models on different races.
Authors
- Bram Ruijsink (ORCID: https://orcid.org/0000-0001-8313-5709)
- Tiarna Lee (ORCID: https://orcid.org/0000-0001-5180-8884)
- Andrew P. King (ORCID: https://orcid.org/0000-0002-9965-7015)
- Theodore Barfoot (ORCID: https://orcid.org/0009-0005-2309-5238)
- Shaheim Ogbomo-Harmitt (ORCID: https://orcid.org/0000-0002-9547-4828)
- Esther Puyol‐Antón (ORCID: https://orcid.org/0000-0002-9789-6629)
- Sébastien Roujol (ORCID: https://orcid.org/0000-0002-7146-0552)
- Miaojing Shi (ORCID: https://orcid.org/0000-0002-4933-0073)
Institutions
- Tongji University (CN)
- St Thomas' Hospital (GB)
- King's College London (GB)
- King's College School (GB)
- St. Thomas Hospital (CA)
- HeartFlow (United States) (US)
Publication Details
- Journal
- European Heart Journal - Digital Health
- Published
- 2025-02-24
- DOI
- https://doi.org/10.1093/ehjdh/ztaf008
- Citations
- 4
- Primary Topic
- Cardiovascular Disease and Adiposity
- Type
- article
- Field-Weighted Citation Impact
- 5.55