Dataset · Silversorrow
Medical VQA benchmark for spine diseases
Silversorrow/SpineBench
64,878 English QA pairs for 40,263 spine images to evaluate multimodal models.
What's inside
SpineBench is a specialized Visual Question Answering (VQA) benchmark for spine disease diagnosis. It contains 64,878 English question-answer pairs linked to 40,263 medical images covering 11 spine pathologies. The dataset is designed for detailed analysis and reasoning evaluation of large multimodal models (MLLM/LVLM) in medicine.
What you can do with it
- Evaluating accuracy and clinical reasoning capabilities of multimodal LLMs in spine image analysis
- Fine-tuning (SFT) medical vision-language models for multiple-choice question answering
- Tasks
- Visual question answering, Multiple choice
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 29.2K
- Likes
- 0
How to load it
from datasets import load_dataset
ds = load_dataset("Silversorrow/SpineBench")