AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Constructing Ophthalmic MLLM for Positioning-diagnosis Collaboration Through Clinical Cognitive Chain Reasoning

Xinyao Liu, Diping Song

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Multimodal LLM benchmarks · 8 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add multi-turn cognitive chain structure instead of single-turn instructions (ablation)FundusGen fine-tuning · GMAI-MMBenchaccuracy: 63.2% -> 66.7%Improved
add multi-turn cognitive chain structure instead of single-turn instructions (ablation)FundusGen fine-tuning · Fundus-MMBenchaccuracy: 67.1% -> 69.7%Improved
add region-annotated instructions (ablation)FundusGen fine-tuning · Fundus-MMBenchaccuracy: 65.3% -> 68.9%Improved
add region-annotated instructions (ablation)FundusGen fine-tuning · GMAI-MMBenchaccuracy: 59.3% -> 64.7%Improved
add General Report startup data (ablation)FundusGen fine-tuning · Fundus-MMBenchaccuracy: 62.9% -> 66.2%Improved
add General Report startup data (ablation)FundusGen fine-tuning · GMAI-MMBenchaccuracy: 56.4% -> 60.9%Improved
add cognitive chain data for 21 complex diseases (ablation)FundusGen fine-tuning · GMAI-MMBench (21 complex diseases)accuracy: 70.4% -> 75.2%Improved
fine-tune on FundusGenLLaVA-v1.5 · Fundus-MMBench and GMAI-MMBench averageaccuracy: 26.9% -> 40.0%Improved
fine-tune on FundusGenQwen2-VL · Fundus-MMBench and GMAI-MMBench averageaccuracy: 34.7% -> 56.9%Improved
fine-tune InternVL2.5 on FundusGenInternVL2.5-8B · GMAI-MMBenchaccuracy: 36.5% -> 66.7%Improved
use FundusGen with cognitive chains instead of Classification Annotation-Guided DataInternVL2.5-8B · GMAI-MMBenchaccuracy vs data scaleImproved
generate synthetic reports with FundusExpert instead of GPT-4oQwen2-VL-7B · Fundus-MMBenchaccuracy: 35.6% -> 40.1%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy