DINOv2 with low-rank adaptation achieved the best performance for classifying carotid atherosclerosis from retinal images with an AUC of 0.71 and significantly predicted future cardiovascular mortality.
Cohort (n=39,620)
No
Do vision foundation models applied to retinal images accurately predict carotid atherosclerosis and future CVD mortality?
Explainable foundation models, particularly DINOv2, demonstrate feasibility for opportunistic screening of carotid atherosclerosis and future CVD mortality using retinal imaging.
Effect estimate: AUC 0.71 (95% CI 0.68-0.73)
Carotid atherosclerosis is a key predictor of cardiovascular disease (CVD), necessitating early detection. While foundation models (FMs) show promise in medical imaging, their optimal selection and fine-tuning strategies for classifying carotid atherosclerosis from retinal images remain unclear. Using data from 39,620 individuals, we evaluated four vision FMs with three fine-tuning methods. Performance was evaluated by predictive performance, clinical utility by survival analysis for future CVD mortality, and explainability by Grad-CAM with vessel segmentation. DINOv2 with low-rank adaptation showed the best overall performance (area under the receiver operating characteristic curve = 0.71; sensitivity = 0.87; specificity = 0.44), prognostic relevance (hazard ratio = 2.20, P-trend < 0.05), and vascular alignment. While further external validation on a broader clinical context is necessary to improve the model's generalizability, these findings support the feasibility of opportunistic atherosclerosis and CVD screening using retinal imaging and highlight the importance of a multi-dimensional evaluation framework for optimal FM selection in medical artificial intelligence.
Lee et al. (Tue,) conducted a cohort in Carotid atherosclerosis (n=39,620). DINOv2 with low-rank adaptation (LoRA) vs. Other foundation models and fine-tuning methods was evaluated on Carotid atherosclerosis classification (AUC 0.71, 95% CI 0.68-0.73). DINOv2 with low-rank adaptation achieved the best performance for classifying carotid atherosclerosis from retinal images with an AUC of 0.71 and significantly predicted future cardiovascular mortality.