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

Area of Science:

• Ophthalmology
• Radiology
• Medical Imaging

Background:

• Thyroid eye disease (TED) is an autoimmune disorder with characteristic orbital magnetic resonance imaging (MRI) findings.
• Accurate segmentation of orbital structures is crucial for TED diagnosis and management.
• Existing orbital MRI datasets lack comprehensive annotations and clinical data.

Purpose of the Study:

• To introduce TOM500, the largest publicly available annotated orbital MRI dataset for thyroid eye disease.
• To provide a valuable resource for developing and validating computational tools for TED analysis.
• To facilitate advancements in TED diagnosis, classification, and monitoring through AI.

Main Methods:

• Compilation of T2-weighted orbital MRI scans from 500 TED patients.
• Annotation of nine key orbital structures by junior annotators, reviewed by an expert radiologist.
• Inclusion of comprehensive clinical data, including demographics and disease duration.

Main Results:

• Creation of TOM500, a multi-organ annotated orbital MRI dataset.
• Evaluation of segmentation consistency using the intraclass correlation coefficient.
• Provision of a dataset suitable for machine learning model development.

Conclusions:

• TOM500 represents a significant contribution to the field of orbital imaging for TED.
• The dataset will accelerate the development of automated diagnostic and classification tools for TED.
• This resource supports future research in computational ophthalmology and radiology.