.png){kind=icon}
.png){kind=icon}
Return
Compact vision language models enable efficient and interpretable optical coherence tomography through layer-specific multimodal learning
.png){kind=icon}
OA
AI
DOI：10.1038/s43856-025-01293-9.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Translating the intricate anatomical signatures of retinal disease from optical coherence tomography (OCT) B-scans into clear, accurate clinical narratives demands algorithms that seamlessly fuse visual features with domain expertise. We curated a multimodal dataset of 40,000 OCT B-scans from public repositories and private clinical cohorts, each paired with expert-validated summaries spanning six conditions: diabetic macular edema, diabetic retinopathy, geographic atrophy, drusen, choroidal neovascularization, and healthy retina. We introduce LO-VLM, a compact (247M parameter) vision-language model (VLM) that infuses anatomical guidance into both encoder and decoder for free-form summary generation and multiclass disease classification. Benchmarking against state-of-the-art RetinaVLM, LLaVA-Med, and a ViT vision only model demonstrates superior performance. In a blinded evaluation by three board certified retina specialists, LO-VLM narratives achieves a mean = 8.5 (standard deviation = 1.15) out of 10, compared to a mean = 5.5 (standard 32 deviation = 1.13) for RetinaVLM (p < 0.0001). In quantitative evaluations, LO-VLM achieves an SBERT similarity of 80.3% and a BERTScore F1 of 71.5%, representing improvements of 8.2% and 28.8% over specialized VLM baselines. For disease classification, LO-VLM reaches 96% accuracy (F1 = 96%), outperforming ViT by 13% and exceeding medical VLM benchmarks by over 62% (p < 0.05). By reconciling interpretability with computational efficiency, LO-VLM establishes a paradigm for efficient AI models in OCT interpretation. Eye diseases such as diabetic retinopathy and macular degeneration can cause vision loss if not detected early. Doctors use a type of imaging called optical coherence tomography (OCT) to examine the layers of the retina, but interpreting the images obtained can be time-consuming and requires specialist training. In this study, we developed LO-VLM, an artificial intelligence system that learns from both images and descriptions written by experts to enable analysis of OCT scans. Using 40,000 examples, it can accurately identify different retinal diseases and generate clear, human readable summaries. LO-VLM achieves high accuracy and produces reports clinicians find useful. This technology could support faster, more consistent eye disease diagnosis and provide interpretation similar to that seen by specialists in routine and remote care settings. Wang et al at introduce a dynamic, network-based tool to predict the risk of hepatic encephalopathy in patients with cirrhosis. The model integrates expert knowledge with data-driven insights to enable early prediction and risk stratification across diverse patient populations.
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
6.3
Papers:
1.8K
Citations:
2.6K
Organization
View more organizations