Improve SpineBench dataset card with paper link, abstract, and comprehensive citation

README.md

@@ -1,15 +1,15 @@
 ---
+language:
+- en
+license: apache-2.0
+size_categories:
+- 10K<n<100K
 task_categories:
 - visual-question-answering
 - multiple-choice
+pretty_name: SpineBench
 tags:
 - medical
-size_categories:
-- 10K<n<100K
-license: apache-2.0
-language:
-- en
-pretty_name: SpineBench
 configs:
 - config_name: default
   data_files:
@@ -18,39 +18,41 @@ configs:
   - split: all
     path: all/train.json
 dataset_info:
-  - name: question_id
+- name: question_id
+  dtype: int64
+- name: split
+  dtype: string
+- name: task
+  dtype: string
+- name: question
+  dtype: string
+- name: options
+  sequence: string
+- name: image
+  dtype: image
+- name: answers
+  sequence: string
+- name: metadata
+  struct:
+  - name: age
     dtype: int64
-  - name: split
-    dtype: string
-  - name: task
+  - name: gender
     dtype: string
-  - name: question
-    dtype: string
-  - name: options
-    sequence: string
-  - name: image
-    dtype: image
-  - name: answers
-    sequence: string
-  - name: metadata       
-    struct:           
-    - name: age            
-      dtype: int64    
-    - name: gender           
-      dtype: string   
 ---
 
-
 ## Dataset Card for SpineBench
 
-- [Benchmark Details](https://huggingface.co/datasets/Silversorrow/SpineBench/blob/main/README.md#benchmark-details)
-- [Paper Information](https://huggingface.co/datasets/Silversorrow/SpineBench/blob/main/README.md#paper-information)
-- [Benchmark Examples](https://huggingface.co/datasets/Silversorrow/SpineBench/blob/main/README.md#benchmark-example)
-- [Benchmark Distribution]((https://huggingface.co/datasets/Silversorrow/SpineBench/blob/main/README.md#benchmark-distribution))
-- [Data Format](https://huggingface.co/datasets/Silversorrow/SpineBench/blob/main/README.md#data-format)
-- [Data Source](https://huggingface.co/datasets/Silversorrow/SpineBench/blob/main/README.md#data-source)
-- [Human Evaluation of MLLMs Reasoning Performance](https://huggingface.co/datasets/Silversorrow/SpineBench/blob/main/README.md#human-evaluation-of-mllms-reasoning-performance)
-- [Citation](https://huggingface.co/datasets/Silversorrow/SpineBench/blob/main/README.md#citation)
+[Paper](https://huggingface.co/papers/2510.12267) | [Project Page](https://zhangchenghanyu.github.io/SpineBench.github.io/)
+
+- [Benchmark Details](#benchmark-details)
+- [Paper Information](#paper-information)
+- [Abstract](#abstract)
+- [Benchmark Examples](#benchmark-example)
+- [Benchmark Distribution](#benchmark-distribution)
+- [Data Format](#data-format)
+- [Data Source](#data-source)
+- [Human Evaluation of MLLMs Reasoning Performance](#human-evaluation-of-mllms-reasoning-performance)
+- [Citation](#citation)
 
 ## Benchmark Details
 
@@ -59,8 +61,14 @@ dataset_info:
 <img src="https://cdn-uploads.huggingface.co/production/uploads/677c80f604c67a51c9716c3b/Udjs8vpxKWwMU2oSWzQQS.png" style="zoom:40%;" />
 ## Paper Information
 
-- Project: https://zhangchenghanyu.github.io/SpineBench.github.io/
-- Appendix: https://zhangchenghanyu.github.io/SpineBench.github.io/#Appendix
+Here you can find more information about the paper and project:
+- **Paper:** [SpineBench: Benchmarking Multimodal LLMs for Spinal Pathology Analysis](https://huggingface.co/papers/2510.12267)
+- **Project Page:** [SpineBench Project](https://zhangchenghanyu.github.io/SpineBench.github.io/)
+- **Appendix:** [SpineBench Appendix](https://zhangchenghanyu.github.io/SpineBench.github.io/#Appendix)
+
+## Abstract
+
+With the increasing integration of Multimodal Large Language Models (MLLMs) into the medical field, comprehensive evaluation of their performance in various medical domains becomes critical. However, existing benchmarks primarily assess general medical tasks, inadequately capturing performance in nuanced areas like the spine, which relies heavily on visual input. To address this, we introduce SpineBench, a comprehensive Visual Question Answering (VQA) benchmark designed for fine-grained analysis and evaluation of MLLMs in the spinal domain. SpineBench comprises 64,878 QA pairs from 40,263 spine images, covering 11 spinal diseases through two critical clinical tasks: spinal disease diagnosis and spinal lesion localization, both in multiple-choice format. SpineBench is built by integrating and standardizing image-label pairs from open-source spinal disease datasets, and samples challenging hard negative options for each VQA pair based on visual similarity (similar but not the same disease), simulating real-world challenging scenarios. We evaluate 12 leading MLLMs on SpineBench. The results reveal that these models exhibit poor performance in spinal tasks, highlighting limitations of current MLLM in the spine domain and guiding future improvements in spinal medicine applications. SpineBench is publicly available at this https URL .
 
 ## Benchmark Example
 
@@ -153,4 +161,12 @@ we invite collaborating clinical physicians to evaluate model reasoning processe
 
 If you use the **SpineBench** dataset in your work, please kindly cite the paper using this BibTeX:
 
-(It will be published on Arxiv soon...)
+```bibtex
+@article{zhang2024spinebench,
+  title={SpineBench: Benchmarking Multimodal LLMs for Spinal Pathology Analysis},
+  author={Zhang, Chenghan and Li, Hongwei and Li, Jing and Wang, Zidong and Lu, Wei and Liu, Kai and Chen, Yuqian and Wang, Zixi and Xiao, Mengting and Xu, Yong and et al.},
+  journal={arXiv preprint arXiv:2510.12267},
+  year={2024},
+  url={https://huggingface.co/papers/2510.12267}
+}
+```