harvardairobotics/FairDomain

---
license: cc-by-nc-nd-4.0
task_categories:
- image-segmentation
- image-classification
modality:
- image
language:
- en
tags:
- medical
- ophthalmology
- fairness
- domain-shift
- fundus
- glaucoma
pretty_name: Harvard-FairDomain
size_categories:
- 10K<n<100K
---

# Dataset Card: Harvard-FairDomain

## Dataset Summary

Harvard-FairDomain is a large-scale ophthalmology dataset designed for studying **fairness under domain shift** in medical image analysis. It supports both image segmentation and classification tasks, with 10,000 samples per task drawn from 10,000 unique patients. The dataset introduces an additional imaging modality — en-face fundus images — alongside the original scanning laser ophthalmoscopy (SLO) fundus images, enabling cross-domain fairness research.

This dataset was introduced in the ECCV 2024 paper: [FairDomain: Achieving Fairness in Cross-Domain Medical Image Segmentation and Classification](https://arxiv.org/pdf/2407.08813).

## Dataset Details

### Dataset Description

| Field | Value |
|-----------------|-------|
| **Institution** | Department of Ophthalmology, Harvard Medical School |
| **Tasks** | Medical image segmentation, medical image classification |
| **Modalities** | En-face fundus image, scanning laser ophthalmoscopy (SLO) fundus image |
| **Samples** | 10,000 (segmentation), 10,000 (classification) |
| **Patients** | 10,000 per task (unique patients) |

### Source Data

Harvard-FairDomain is derived from two existing Harvard ophthalmology datasets:

- [**Harvard-FairSeg**](https://github.com/Harvard-Ophthalmology-AI-Lab/FairSeg) — source for segmentation task data
- [**FairVLMed (FairCLIP)**](https://github.com/Harvard-Ophthalmology-AI-Lab/FairCLIP) — source for classification task data

En-face fundus images were added to both subsets as a new imaging domain on top of the original SLO fundus images, enabling cross-domain fairness benchmarking.

## Uses

### Direct Use

Research on algorithmic fairness in cross-domain medical image segmentation and classification, including studies of model performance disparities across demographic groups under distribution shift.

### Out-of-Scope Use

Clinical diagnosis, commercial applications, or any use prohibited by the CC BY-NC-ND 4.0 license.

## Citation

**BibTeX:**

```bibtex
@article{tian2024fairdomain,
title={FairDomain: Achieving Fairness in Cross-Domain Medical Image Segmentation and Classification},
author={Tian, Yu and Wen, Congcong and Shi, Min and Afzal, Muhammad Muneeb and Huang, Hao and Khan, Muhammad Osama and Luo, Yan and Fang, Yi and Wang, Mengyu},
journal={arXiv preprint arXiv:2407.08813},
year={2024}
}
```

**APA:**

Tian, Y., Wen, C., Shi, M., Afzal, M. M., Huang, H., Khan, M. O., Luo, Y., Fang, Y., & Wang, M. (2024). FairDomain: Achieving Fairness in Cross-Domain Medical Image Segmentation and Classification. *arXiv preprint arXiv:2407.08813*.