Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone images

BackgroundThe development of medical artificial intelligence (AI) models is primarily driven by the need to address healthcare resource scarcity, particularly in underserved regions. Proposing an affordable, accessible, interpretable, and automated AI system for non-clinical settings is crucial to e...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhongwen Li, Zhouqian Wang, Liheng Xiu, Pengyao Zhang, Wenfang Wang, Yangyang Wang, Gang Chen, Weihua Yang, Wei Chen
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-05-01
Series:Frontiers in Cell and Developmental Biology
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fcell.2025.1600202/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849309933167706112
author Zhongwen Li
Zhongwen Li
Zhouqian Wang
Liheng Xiu
Pengyao Zhang
Wenfang Wang
Yangyang Wang
Gang Chen
Weihua Yang
Wei Chen
author_facet Zhongwen Li
Zhongwen Li
Zhouqian Wang
Liheng Xiu
Pengyao Zhang
Wenfang Wang
Yangyang Wang
Gang Chen
Weihua Yang
Wei Chen
author_sort Zhongwen Li
collection DOAJ
description BackgroundThe development of medical artificial intelligence (AI) models is primarily driven by the need to address healthcare resource scarcity, particularly in underserved regions. Proposing an affordable, accessible, interpretable, and automated AI system for non-clinical settings is crucial to expanding access to quality healthcare.MethodsThis cross-sectional study developed the Multimodal Ocular Surface Assessment and Interpretation Copilot (MOSAIC) using three multimodal large language models: gpt-4-turbo, claude-3-opus, and gemini-1.5-pro-latest, for detecting three ocular surface diseases (OSDs) and grading keratitis and pterygium. A total of 375 smartphone-captured ocular surface images collected from 290 eyes were utilized to validate MOSAIC. The performance of MOSAIC was evaluated in both zero-shot and few-shot settings, with tasks including image quality control, OSD detection, analysis of the severity of keratitis, and pterygium grading. The interpretability of the system was also evaluated.ResultsMOSAIC achieved 95.00% accuracy in image quality control, 86.96% in OSD detection, 88.33% in distinguishing mild from severe keratitis, and 66.67% in determining pterygium grades with five-shot settings. The performance significantly improved with the increasing learning shots (p < 0.01). The system attained high ROUGE-L F1 scores of 0.70–0.78, depicting its interpretable image comprehension capability.ConclusionMOSAIC exhibited exceptional few-shot learning capabilities, achieving high accuracy in OSD management with minimal training examples. This system has significant potential for smartphone integration to enhance the accessibility and effectiveness of OSD detection and grading in resource-limited settings.
format Article
id doaj-art-09e2ec12ddfc46fdbdaa73990e7c2dea
institution Kabale University
issn 2296-634X
language English
publishDate 2025-05-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Cell and Developmental Biology
spelling doaj-art-09e2ec12ddfc46fdbdaa73990e7c2dea2025-08-20T03:53:56ZengFrontiers Media S.A.Frontiers in Cell and Developmental Biology2296-634X2025-05-011310.3389/fcell.2025.16002021600202Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone imagesZhongwen Li0Zhongwen Li1Zhouqian Wang2Liheng Xiu3Pengyao Zhang4Wenfang Wang5Yangyang Wang6Gang Chen7Weihua Yang8Wei Chen9Ningbo Key Laboratory of Medical Research on Blinding Eye Diseases, Ningbo Eye Institute, Ningbo Eye Hospital, Wenzhou Medical University, Ningbo, ChinaNational Clinical Research Center for Ocular Diseases, Eye Hospital, Wenzhou Medical University, Wenzhou, ChinaNingbo Key Laboratory of Medical Research on Blinding Eye Diseases, Ningbo Eye Institute, Ningbo Eye Hospital, Wenzhou Medical University, Ningbo, ChinaDepartment of Ophthalmology, West China Second University Hospital, Sichuan University, Chengdu, ChinaNingbo Key Laboratory of Medical Research on Blinding Eye Diseases, Ningbo Eye Institute, Ningbo Eye Hospital, Wenzhou Medical University, Ningbo, ChinaNingbo Key Laboratory of Medical Research on Blinding Eye Diseases, Ningbo Eye Institute, Ningbo Eye Hospital, Wenzhou Medical University, Ningbo, ChinaNingbo Key Laboratory of Medical Research on Blinding Eye Diseases, Ningbo Eye Institute, Ningbo Eye Hospital, Wenzhou Medical University, Ningbo, ChinaFirst People’s Hospital of Aksu, Aksu, ChinaShenzhen Eye Hospital, Shenzhen Eye Medical Center, Southern Medical University, Shenzhen, ChinaNational Clinical Research Center for Ocular Diseases, Eye Hospital, Wenzhou Medical University, Wenzhou, ChinaBackgroundThe development of medical artificial intelligence (AI) models is primarily driven by the need to address healthcare resource scarcity, particularly in underserved regions. Proposing an affordable, accessible, interpretable, and automated AI system for non-clinical settings is crucial to expanding access to quality healthcare.MethodsThis cross-sectional study developed the Multimodal Ocular Surface Assessment and Interpretation Copilot (MOSAIC) using three multimodal large language models: gpt-4-turbo, claude-3-opus, and gemini-1.5-pro-latest, for detecting three ocular surface diseases (OSDs) and grading keratitis and pterygium. A total of 375 smartphone-captured ocular surface images collected from 290 eyes were utilized to validate MOSAIC. The performance of MOSAIC was evaluated in both zero-shot and few-shot settings, with tasks including image quality control, OSD detection, analysis of the severity of keratitis, and pterygium grading. The interpretability of the system was also evaluated.ResultsMOSAIC achieved 95.00% accuracy in image quality control, 86.96% in OSD detection, 88.33% in distinguishing mild from severe keratitis, and 66.67% in determining pterygium grades with five-shot settings. The performance significantly improved with the increasing learning shots (p < 0.01). The system attained high ROUGE-L F1 scores of 0.70–0.78, depicting its interpretable image comprehension capability.ConclusionMOSAIC exhibited exceptional few-shot learning capabilities, achieving high accuracy in OSD management with minimal training examples. This system has significant potential for smartphone integration to enhance the accessibility and effectiveness of OSD detection and grading in resource-limited settings.https://www.frontiersin.org/articles/10.3389/fcell.2025.1600202/fullocular surface diseaselarge language modelmultimodal modelkeratitisconjunctivitispterygium
spellingShingle Zhongwen Li
Zhongwen Li
Zhouqian Wang
Liheng Xiu
Pengyao Zhang
Wenfang Wang
Yangyang Wang
Gang Chen
Weihua Yang
Wei Chen
Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone images
Frontiers in Cell and Developmental Biology
ocular surface disease
large language model
multimodal model
keratitis
conjunctivitis
pterygium
title Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone images
title_full Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone images
title_fullStr Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone images
title_full_unstemmed Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone images
title_short Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone images
title_sort large language model based multimodal system for detecting and grading ocular surface diseases from smartphone images
topic ocular surface disease
large language model
multimodal model
keratitis
conjunctivitis
pterygium
url https://www.frontiersin.org/articles/10.3389/fcell.2025.1600202/full
work_keys_str_mv AT zhongwenli largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT zhongwenli largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT zhouqianwang largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT lihengxiu largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT pengyaozhang largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT wenfangwang largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT yangyangwang largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT gangchen largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT weihuayang largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages
AT weichen largelanguagemodelbasedmultimodalsystemfordetectingandgradingocularsurfacediseasesfromsmartphoneimages