OpenMed/synthvision-seeds

---
license: apache-2.0
task_categories:
- visual-question-answering
tags:
- medical
- synthvision
- openmed
size_categories:
- 100K<n<1M
---

# synthvision-seeds


Seed records from 4 open medical image datasets

**Records**: 119,137

## About

Seed dataset for the [SynthVision pipeline](https://huggingface.co/blog/OpenMed/synthvision). Contains 119,137 records aggregated from 4 open medical image datasets:

| Source | Records | Modality |
|--------|---------|----------|
| [eltorio/ROCO-radiology](https://huggingface.co/datasets/eltorio/ROCO-radiology) | 65,393 | Radiology |
| [OpenMed/multicare-images](https://huggingface.co/datasets/OpenMed/multicare-images) | 50,000 | Mixed |
| [flaviagiammarino/path-vqa](https://huggingface.co/datasets/flaviagiammarino/path-vqa) | 3,430 | Pathology |
| [flaviagiammarino/vqa-rad](https://huggingface.co/datasets/flaviagiammarino/vqa-rad) | 314 | Radiology |

Images are deduplicated by SHA-256 hash. Each record contains an image path, source dataset ID, modality, and any available metadata (captions or Q&A pairs).

## Schema

```
id: str # unique record ID
image: str # relative image path
source: str # source dataset name
modality: str # imaging modality
metadata: dict # captions, Q&A pairs, or labels
```

## Loading

```python
from datasets import load_dataset

ds = load_dataset("OpenMed/synthvision-seeds")
```

## Links

- [SynthVision blog post](https://huggingface.co/blog/OpenMed/synthvision)
- [Source code](https://github.com/openmed-labs/synthvision)
- [All SynthVision artifacts](https://huggingface.co/collections/OpenMed/synthvision-69baac655b557943aa1babd3)
- [OpenMed on Hugging Face](https://huggingface.co/OpenMed)