Multimodal LLMs for retinal disease diagnosis via OCT: few-shot versus single-shot learning.
利用多模態大型語言模型(LLMs)透過OCT進行視網膜疾病診斷:少量學習(few-shot)與單次學習(single-shot)的比較
Ther Adv Ophthalmol 2025-05-22
這項研究發現,GPT-4o 和 Claude Sonnet 3.5 這兩款AI模型,經過少量範例訓練後,診斷OCT影像的準確率最高可達73%。雖然還不如專業深度學習模型,但在日常眼科診斷、特別是判斷正常個案時,已展現輔助潛力。未來需更多研究結合影像和臨床資料來提升表現。
PubMedDOI♡
站上相關主題文章列表
Visual-textual integration in LLMs for medical diagnosis: A preliminary quantitative analysis.
大型語言模型在醫學診斷中的視覺-文本整合:初步定量分析。
Comput Struct Biotechnol J 2025-01-24
Multimodal machine learning enables AI chatbot to diagnose ophthalmic diseases and provide high-quality medical responses.
多模態機器學習使 AI 聊天機器人能夠診斷眼科疾病並提供高品質的醫療回應。
NPJ Digit Med 2025-01-27
Can off-the-shelf visual large language models detect and diagnose ocular diseases from retinal photographs?
現成的視覺大型語言模型能否從視網膜照片中檢測和診斷眼科疾病?
BMJ Open Ophthalmol 2025-04-07
Diagnostic performance of multimodal large language models in radiological quiz cases: the effects of prompt engineering and input conditions.
多模態大型語言模型於放射學問答案例中的診斷表現:提示工程與輸入條件的影響
Ultrasonography 2025-04-16
這項研究比較三款多模態大型語言模型在解讀放射影像的表現,發現 Claude 3.5 Sonnet 準確率最高。使用 AI 生成的提示語和在影像中加上描述文字,都能明顯提升診斷效果。模型表現會受病例罕見度和知識截止日影響。整體來說,善用提示工程和豐富輸入資料,有助提升 LLMs 在放射科的應用效能。
PubMedDOI
Multimodal Performance of GPT-4 in Complex Ophthalmology Cases.
GPT-4 在複雜眼科病例中的多模態表現
J Pers Med 2025-04-25
Image Recognition Performance of GPT-4V(ision) and GPT-4o in Ophthalmology: Use of Images in Clinical Questions.
GPT-4V(ision) 與 GPT-4o 在眼科影像辨識的表現:臨床問題中影像應用之探討
Clin Ophthalmol 2025-05-13
Diagnostic Performance of Publicly Available Large Language Models in Corneal Diseases: A Comparison with Human Specialists.
公開大型語言模型在角膜疾病診斷表現之評估:與人類專科醫師的比較
Diagnostics (Basel) 2025-05-28
Performance of a novel multimodal large language model in ınterpreting meibomian glands quantitatively and qualitatively.
新型多模態大型語言模型於定量與定性解讀瞼板腺的表現
Int Ophthalmol 2025-05-28
Claude 3.5 Sonnet 是多模態大型語言模型,這次用來分析160張瞼板腺攝影,評分腺體脫落程度並偵測形態異常。它在腺體流失分級的準確率高達85–97.5%,能正確辨識所有人工標記的異常,還能發現人工漏掉的細微問題,顯示有潛力成為眼科影像判讀和教學的好幫手。
PubMedDOI
Multi-Modal AI for Multi-Label Retinal Disease Prediction Using OCT and Fundus Images: A Hybrid Approach.
多模態 AI 結合 OCT 與眼底影像於多標籤視網膜疾病預測之混合式方法
Sensors (Basel) 2025-07-30
這篇論文提出 VisionTrack 多模態 AI 系統,能整合影像、臨床資料和醫療報告,提升視網膜疾病診斷的準確度。系統結合 CNN、GNN 和 LLM 技術,在公開資料集上測試表現優異,展現早期偵測和個人化眼科照護的應用潛力。
PubMedDOI