TOM500: A Multi-Organ Annotated Orbital MRI Dataset for Thyroid Eye Disease

Abstract This study presents TOM500, a comprehensive multi-organ annotated orbital magnetic resonance imaging (MRI) dataset. It includes clinical data, T2-weighted MRI scans, and corresponding segmentations from 500 patients with thyroid eye disease (TED) during their initial visit. TED is a common autoimmune disorder with distinct orbital MRI features. Segmentations of nine orbital structures, including the optic nerve, orbital fat, lacrimal gland, eyeball, and five extraocular muscles (superior rectus and levator palpebrae superioris complex, inferior rectus, medial rectus, lateral rectus, and superior oblique), were generated by three junior annotators and reviewed by an expert radiologist. The consistency of the segmentations was evaluated using the intraclass correlation coefficient. Clinical data, including sex, age, disease duration, and smoking status, are also provided for disease diagnosis and classification. TOM500, the largest publicly available orbital MRI dataset with expert annotations, is designed to facilitate the development of advanced computational tools for TED diagnosis, classification, and observation.