National Multicenter Study Using AI and Orbital MRI to Diagnose and Predict Outcomes in Thyroid Eye Disease

What this Trial Is About

This research aims to validate the effectiveness of an artificial intelligence AI multimodal model using multi-sequence orbital MRI to diagnose, assess activity and severity, and predict the prognosis of thyroid-associated ophthalmopathy TAO in real-world clinical settings. It focuses on whether AI can accurately detect the presence and stage of TAO and predict patient outcomes by comparing AI performance between TAO patients and healthy individuals. Participants will receive standardized multimodal orbital MRI scans using various sequences T1WI, T2WI, STIR, and research sequences like Magic, IDEAL-IQ, DWI, ASL, CEST. Ocular ultrasound images will also be collected from TAO patients at different disease stages, ophthalmic disease controls, and healthy volunteers. AI deep learning techniques will automatically segment key eye structures and extract detailed imaging features to build prediction models for TAO screening and activity staging, validated against MRI findings and clinical activity scores. Throughout the study, participants will undergo imaging assessments and clinical evaluations over an 18-month follow-up period. Researchers will measure diagnostic accuracy, classification accuracy for disease activity and severity, and predictive performance of the AI model. They will also assess agreement between AI and clinical experts, correlations between imaging and pathology, and AIs ability to predict treatment response using baseline and follow-up imaging data.

Screening, Classification, and Outcome Prediction Ensemble Using Orbital MRI for Thyroid Eye Disease: The TED SCOPE National Multicenter Registry Study