Academic Journal
Parameter-efficient contrastive language-image pre-training (CLIP) adaptation for few-shot anomaly detection in ultra-widefield fundus images.
| Title: | Parameter-efficient contrastive language-image pre-training (CLIP) adaptation for few-shot anomaly detection in ultra-widefield fundus images. |
|---|---|
| Authors: | Liao G; College of Biomedical Engineering, Sichuan University, Chengdu 610065, China. Electronic address: leosigmoid@stu.scu.edu.cn., Xu H; Department of Ophthalmology, West China Hospital, Chengdu 610065, China. Electronic address: XuHanyue@wchscu.cn., Ai H; College of Biomedical Engineering, Sichuan University, Chengdu 610065, China. Electronic address: aihaoyue@stu.scu.edu.cn., Wang T; Qingdao Eye Hospital of Shandong First Medical University, Qingdao 266071, China. Electronic address: tonewang108@163.com., Huang Y; Department of Ophthalmology, West China Hospital, Chengdu 610065, China. Electronic address: huangyufan0606@163.com., Han L; College of Biomedical Engineering, Sichuan University, Chengdu 610065, China. Electronic address: gjzzcc@163.com., Zhuang Y; College of Biomedical Engineering, Sichuan University, Chengdu 610065, China. Electronic address: zhuangy@scu.edu.cn., Chen K; College of Biomedical Engineering, Sichuan University, Chengdu 610065, China. Electronic address: chenke@scu.edu.cn., Zhang M; Department of Ophthalmology, West China Hospital, Chengdu 610065, China. Electronic address: zhangmingscu0905@163.com., Lin J; College of Biomedical Engineering, Sichuan University, Chengdu 610065, China. Electronic address: linjiangli@scu.edu.cn. |
| Source: | Photodiagnosis and photodynamic therapy [Photodiagnosis Photodyn Ther] 2026 Jun; Vol. 59, pp. 105461. Date of Electronic Publication: 2026 Apr 09. |
| Publication Type: | Journal Article |
| Language: | English |
| Journal Info: | Publisher: Elsevier Country of Publication: Netherlands NLM ID: 101226123 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1873-1597 (Electronic) Linking ISSN: 15721000 NLM ISO Abbreviation: Photodiagnosis Photodyn Ther Subsets: MEDLINE |
| Imprint Name(s): | Original Publication: Amsterdam ; Boston : Elsevier, c2004- |
| MeSH Terms: | Image Processing, Computer-Assisted*/methods , Fundus Oculi* , Deep Learning*, Image Interpretation, Computer-Assisted/methods ; Humans ; Algorithms |
| Abstract: | Fundus diseases are among the leading causes of visual impairment and blindness, many of which could be prevented through early intervention. Ultra-widefield fundus (UWF) imaging, offering a wide field of view and enabling non-mydriatic acquisition, has emerged as an ideal modality for screening; however, the volume of cases far exceeds the diagnostic capacity of ophthalmologists, necessitating an automated diagnostic system. Nevertheless, existing deep learning algorithms often encounter challenges including high computational costs, large data requirements, and performance degradation on out-of-distribution data. Domain-adaptive fine-tuning of existing multimodal large models offers a promising pathway to address these limitations. Here, we developed an anomaly detection framework for UWF images based on CLIP. First, we proposed a dual-pathway detection mechanism for accurate anomaly detection. Second, a lightweight hybrid low-rank adapter is designed for parameter-efficient fine-tuning to bridge the domain gap between natural images and fundus images. Moreover, to address the scarcity and variability of abnormal samples, we developed an anomaly synthesis algorithm based on Poisson-blending to broaden and generalize the anomaly paradigm. With only 32 pairs of training samples, the framework achieved efficient and robust anomaly detection. The framework was evaluated on 8 independent cross-source datasets (29,739 images covering >50 anomaly patterns), yielding an average AUC of 85.93% and a peak AUC of 94.54%. Owing to its low computational and data requirements, this framework provides a practical and scalable solution for fundus disease screening and shows strong potential for clinical translation. Code and details are available through https://github.com/JohnLeo-XJTU/AnomalyUWF. (Copyright © 2026. Published by Elsevier B.V.) |
| Competing Interests: | Declaration of competing interest The authors declare no conflicts of interest. |
| Contributed Indexing: | Keywords: Artificial intelligence; Few-shot anomaly detection; Fundus diseases; Ultra-widefield fundus images; Vision-language model |
| Entry Date(s): | Date Created: 20260412 Date Completed: 20260613 Latest Revision: 20260613 |
| Update Code: | 20260615 |
| DOI: | 10.1016/j.pdpdt.2026.105461 |
| PMID: | 41966507 |
| Database: | MEDLINE |
| ISSN: | 1873-1597 |
|---|---|
| DOI: | 10.1016/j.pdpdt.2026.105461 |