AS-OCT dataset with anatomical structure segmentation and scleral spur localization in cataract and glaucoma

Abstract

Anterior Segment Optical Coherence Tomography (AS-OCT) provides high-resolution, non-invasive visualization of the anterior eye and is widely used for clinical diagnosis and surgical planning. However, automated analysis of AS-OCT images remains limited by the lack of publicly available datasets with comprehensive anatomical annotations and disease labels. Here we present the AS-OCT Multi-Structure and Disease Classification Dataset (ASOCT-MSDC), a curated dataset designed for anatomical structure segmentation and disease classification. The dataset contains 1106 AS-OCT images from 1106 eyes of 627 individuals across four clinical categories: normal, glaucoma, cataract, and glaucoma-cataract comorbidity. Each image underwent stringent quality assessment, and three standardized image quality scores—eyelid obscuration, black-line anomaly, and central light artifact—are provided as metadata. Expert-validated annotations include segmentation masks for the anterior chamber, iris, lens, and nucleus, together with scleral spur localization points. By combining anatomical annotations, disease labels, and image quality metadata, ASOCT-MSDC supports research on anatomical segmentation, disease classification, and image quality assessment in anterior segment OCT images.