OCTA-SVR-100K

Version 1
Description

OCTA-100K-SVR is a large-scale synthetic dataset designed for training and evaluating Vision-Language Models (VLMs) on Optical Coherence Tomography Angiography (OCTA) analysis, particularly for explainable diabetic retinopathy (DR) staging. The dataset contains 100,000 pairs of synthetic OCTA images and structured reasoning texts, generated through a pathology-aware simulation pipeline.

Keywords
Conditions
License

No license available

OCTA-100K-SVR is a large-scale synthetic dataset designed for training and evaluating Vision-Language Models (VLMs) on Optical Coherence Tomography Angiography (OCTA) analysis, particularly for explainable diabetic retinopathy (DR) staging. The dataset contains 100,000 pairs of synthetic OCTA images and structured reasoning texts, generated through a pathology-aware simulation pipeline.