Med-Banana-50K: 50K Images for Text-Guided Medical Image Editing

Med-Banana-50K is a large-scale dataset containing 50,000 images for instruction-based medical image editing. It spans three imaging modalities (Chest X-ray, Brain MRI, Fundus Photography) and 23 disease types, featuring bidirectional edits generated from real medical images using the Gemini-2.5-Flash-Image model. The dataset was created by RichardChenZH and was last updated on June 1, 2026.

Use Cases

• Training text-to-image models for medical image editing based on the described instruction-based framework.
• Benchmarking generative models on lesion addition and removal tasks across the three specified medical imaging modalities.
• Developing and evaluating cross-modality medical image synthesis techniques using the dataset's Chest X-ray, Brain MRI, and Fundus Photography examples.
• Studying the performance of AI models on a variety of disease types, as the dataset covers 23 different conditions.

Strengths

• Contains 50,000 images, providing a substantial scale for model training.
• Covers three distinct medical imaging modalities: Chest X-ray, Brain MRI, and Fundus Photography.
• Includes edits for 23 different disease types, offering diversity in medical conditions.
• Features bidirectional edits (lesion addition and removal) for each case, enabling comparative studies.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the source real medical images and the generative model used for editing.

Provenance

Source

huggingface

Collection Method

Edits were generated from real medical images using the Gemini-2.5-Flash-Image model.

Freshness

Last updated 2026-06-01 08:39:32; freshness should be verified.