Related Experiment Video
Updated: May 5, 2026
03:55
Computer-Aided Three-Dimensional Visualization in the Treatment of Locally Advanced Thyroid Cancer
Published on: June 9, 2023
937
Developing a thyroid cancer differentiation state classification system using deep residual networks and metabolic
Yanzhi Zhang1,2, Xiaoxue Du1,2, Sijia Cai1,2
1Department of Head and Neck Surgery, Fudan University Shanghai Cancer Center, Shanghai, People's Republic of China.
NPJ Digital Medicine
|September 24, 2025
Summary
We developed an interpretable deep learning model to classify thyroid cancer progression using multiomic data. This approach identifies key metabolic signatures for early diagnosis and improved clinical decisions.
Area of Science:
- Oncology
- Metabolomics
- Genomics
- Bioinformatics
Background:
- Thyroid cancer exhibits diverse differentiation states, from well-differentiated to anaplastic.
- Metabolic reprogramming is a hallmark of cancer, significantly impacting tumor progression and aggressiveness.
Purpose of the Study:
- To develop an accurate and interpretable deep learning framework for classifying thyroid cancer differentiation states.
- To identify key metabolic signatures associated with thyroid cancer progression using multiomic data.
Main Methods:
- Integrated untargeted metabolomic, whole-exome sequencing, and transcriptomic data from 158 thyroid tumors and 57 matched normal tissues.
- Employed a deep residual network (ResNet) architecture for classification and Shapley additive explanations for interpretability.
- Analyzed single-cell RNA sequencing data to map metabolic reprogramming pathways in dedifferentiated thyroid cancer.
Main Results:
- Developed a 10-gene metabolic signature for pan-pathological classification of thyroid carcinomas.
- Created a 10-metabolite classification model using ResNet, demonstrating direct pathophysiological responsiveness.
- Identified critical metabolic signatures driving thyroid cancer differentiation states through interpretability analysis.
Conclusions:
- Metabolic shifts are fundamental to thyroid cancer progression.
- The proposed interpretable model accurately classifies thyroid cancer differentiation states.
- This framework may facilitate early diagnosis and inform clinical decision-making for thyroid cancer patients.
