Related Experiment Video
Updated: May 7, 2026
05:41
Author Spotlight: Integrating Ultrasound Imaging with Biochemical Markers for Thyroid Disease Diagnosis
Published on: February 9, 2024
1.3K
MRI-based automated segmentation and multiparametric quantification for assessing orbital soft tissue involvement in
Linhan Zhai1, Yan Zeng2, Yu Chen1
1Department of Radiology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.
European Radiology
|May 6, 2026
Summary
A new deep learning framework, TED-Net, accurately segments orbital tissues and quantifies imaging biomarkers for thyroid eye disease (TED) grading. This tool aids in precise assessment of TED severity using MRI data.
Area of Science:
- Ophthalmology
- Radiology
- Artificial Intelligence
Background:
- Thyroid eye disease (TED) assessment relies on MRI, but quantitative metrics for orbital involvement are challenging to extract.
- Current methods lack efficient tools for rapid and accurate parameter extraction from orbital MRI.
Purpose of the Study:
- To develop a deep learning-based multimodal framework for automated segmentation of orbital soft tissues.
- To identify quantitative imaging biomarkers for precise grading of thyroid eye disease (TED).
Main Methods:
- Developed TED-Net, a deep learning model integrating ConvNeXt and Transformer architectures, for segmenting orbital structures.
- Utilized 3T MRI with water-fat separation and FS T2 mapping sequences from 330 TED patients.
- Extracted morphological (volume) and functional (water/fat fraction, T2 relaxation time) parameters.
Main Results:
- TED-Net achieved high segmentation accuracy (Dice > 0.80) across all orbital structures.
- Significant differences in morphological and functional parameters were found between mild and moderate-to-severe TED (p < 0.05).
- The combined volumetric-functional model showed superior diagnostic accuracy (AUC = 0.982) compared to the volumetric model alone (AUC = 0.908).
Conclusions:
- TED-Net enables accurate automated segmentation and multiparametric quantification of orbital soft tissues.
- The framework provides reliable imaging biomarkers for objective assessment of TED severity.
- Clinical implementation of TED-Net facilitates more accurate grading of thyroid eye disease.
