Skip to content

Dataset · UCSC-VLAA

MedTrinity-25M: medical images

UCSC-VLAA/MedTrinity-25M

25M+ medical images across 10 modalities with multimodal annotations for 65 diseases.

What's inside

MedTrinity-25M is a large-scale multimodal medical dataset featuring over 25 million images across 10 modalities. Annotations include global information—disease/lesion type, modality, region description, and region relationships—and local ROI descriptions. The language is English; modalities are image and text; the task is question-answering.

What you can do with it

  • Train medical vision-language models
  • Fine-tune VLMs for ROI description
  • Benchmark multimodal QA across 65 diseases
Tasks
Question answering
Languages
English
Size
10M–100M rows
Downloads this month
4.2K
Likes
216
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("UCSC-VLAA/MedTrinity-25M")