MieDB-100k: A Comprehensive Dataset for Medical Image Editing

📄 Introduction

MieDB-100k is a large-scale, high-quality and diverse dataset for text-guided medical image editing, which includes 112, 228 editing data, covering 69 distinct editing targets and 10 diverse medical image modalities. We categorize editing tasks into three types: Perception, Modification and Transformation, which consider both model's intrinsic understanding and generation abilities on medical images. The dataset is constructed by both modality-specific expert models and rule-based data synthetic methods. Additionally, for some complex tasks such as lesion modification, we introduce individuals with medical knowledge to perform manual quality checks on the data to ensure data quality.

⚙️ Dataset Setup

1. Download compressed MieDB-100k dataset
2. Extract compressed file via:

Benchmark split:

mkdir dataBenchmark
pv dataBenchmark_*.tar | tar -xf - -C dataBenchmark --skip-old-files

Train split:

mkdir dataTrain
pv dataTrain_*.tar | tar -xf - -C dataTrain --skip-old-files

Note:

• pv is used for progress visualization. You can switch to cat if you want to extract in silence manner.
• macOS doesn't support --skip-old-files, use tar -xkf - -C /path/to/dst/ instead after the pipe.

🐑 Citation

@article{miedb100k,
      title={MieDB-100k: A Comprehensive Dataset for Medical Image Editing}, 
      author={Yongfan Lai and Wen Qian and Bo Liu and Hongyan Li and Hao Luo and Fan Wang and Bohan Zhuang and Shenda Hong},
      year={2026},
      journel={Preprint at arXiv}
      url={https://arxiv.org/abs/2602.09587}, 
}