arXiv cs.AI· Gujie Shao, Zixun Xie, Xuechun Xing, Ruixiang Wang, Ziyun Lan, Yanlin Qi, Gangyi Zhang, Yuxin Yang, Dawei Li, Haiming Tang·· 13 小时前AI 评分27
EyeVQA：眼科视觉语言模型基准，从疾病识别到空间定位
EyeVQA: Benchmarking Ophthalmic Vision-Language Models from Recognition to Spatial Grounding
AI 导读
研究团队构建了 EyeVQA 眼科视觉问答基准，整合 21 个眼科数据集、2 万条问答对，覆盖 6 类疾病和 7 种题型，其中 44.5% 的题目需跨多张图像推理。在统一零样本协议下评测 14 个通用、科学与医学专用 VLM，最佳模型总分仅 62.8，空间定位与跨任务泛化差距明显。
正文 · 原文
Abstract:Vision-language models (VLMs) have shown increasing potential for medical image understanding, yet their capabilities in ophthalmic imaging remain insufficiently characterized. Existing ophthalmic datasets are typically designed for individual diseases or specialized tasks, making it difficult to systematically evaluate whether VLMs can move beyond disease recognition toward comparative reasoning and fine-grained spatial grounding. We introduce EyeVQA, a unified visual question answering benchmark for comprehensive evaluation of ophthalmic VLMs. EyeVQA is constructed from 21 available ophthalmic datasets and contains 20,000 question-answer pairs spanning six disease groups and seven question types: Single-Choice, Multi-Select, Variable-Select, True-False, Ranking, Point Location, and Bounding Box. Gold answers are deterministically derived from source-provided diagnoses, severity grades, clinical findings, segmentation masks, bounding boxes, and anatomical landmarks, enabling reproducible evaluation without relying on model-generated annotations. Notably, 44.5% of the questions require reasoning across multiple images, extending evaluation beyond conventional single-image medical VQA. We benchmark fourteen representative general-purpose, scientific, and medically specialized VLMs under a unified zero-shot protocol. The best-performing model only achieves an overall score of 62.8, while substantial gaps remain in spatial grounding and cross-task generalization. These results highlight the limitations of current VLMs in comprehensive ophthalmic visual understanding and establish EyeVQA as a diagnostic benchmark for developing more reliable and spatially grounded ophthalmic multimodal models. The project page is available at this https URL.
| Comments: | Accepted to MICCAI 2026 CREATE Workshop. 19 pages, 2 figures, 2 tables. Project page: this https URL |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.32352 [cs.CV] |
| (or arXiv:2609.32352v3 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2609.32352 arXiv-issued DOI via DataCite |
Submission history
From: Haiming Tang [view email]
[v1]
Sat, 26 Sep 2026 08:15:17 UTC (2,126 KB)
[v2]
Tue, 29 Sep 2026 01:27:10 UTC (2,126 KB)
[v3]
Tue, 20 Oct 2026 21:56:44 UTC (2,126 KB)
来源：arXiv cs.AI · arxiv.org