Related Experiment Video
Updated: Aug 5, 2026
05:41
Synchronous Triplanar Reconstruction Integrated with Color Doppler Mapping for Precise and Rapid Localization of Thyroid Lesions
Published on: February 9, 2024
Deep learning models for predicting clinical activity and characteristics in thyroid eye disease using magnetic
Sarada Sriya Rajyam1, Yonca O Arat1, Shoaib Ugradar2
1Department of Ophthalmology and Visual Sciences, Yale School of Medicine, New Haven, Connecticut, USA.
Orbit (Amsterdam, Netherlands)
|July 28, 2026
Summary
Deep learning models predict thyroid eye disease activity and patient traits using orbital MRI scans. This AI approach offers granular disease assessment from imaging data.
Area of Science:
- Ophthalmology
- Radiology
- Artificial Intelligence
Background:
- Thyroid eye disease (TED) requires accurate assessment of disease activity and patient characteristics.
- Orbital magnetic resonance imaging (MRI) provides detailed anatomical information but extracting clinical data can be challenging.
Purpose of the Study:
- To develop and evaluate deep learning models for predicting continuous clinical activity score and key patient characteristics in TED using orbital MRI.
- To demonstrate the feasibility of extracting clinical information directly from orbital MRI scans.
Main Methods:
- Utilized the TOM500 dataset of 500 TED patients' orbital MRI scans and clinical data.
- Adapted a ResNet-50 convolutional neural network for predicting clinical activity score, age, smoking status, and sex.
- Preprocessed MRI slices into 3-channel inputs (raw scan, segmentation mask, weighted average) for model training.
Main Results:
- The clinical activity score prediction model achieved a mean absolute error (MAE) of 1.05.
- Age prediction yielded an MAE of 6.90 years.
- Sex and smoking status prediction achieved accuracies of 93% and 80%, respectively, with areas under the ROC curve of 0.974 and 0.747.
Conclusions:
- Artificial intelligence, specifically deep learning, shows significant promise in analyzing orbital MRI for TED.
- This AI-driven approach enables granular assessment of disease progression and extraction of patient characteristics directly from orbital MRI.
