Related Experiment Video
Updated: May 10, 2025
10:01
Modeling Brain Metastasis by Internal Carotid Artery Injection of Cancer Cells
Published on: August 2, 2022
6.1K
Differentiation of Glioblastoma and Solitary Brain Metastasis Using Brain-Tumor Interface Radiomics Features Based on
Yini Chen1, Weiya Guo2, Yushi Li3
1Department of Radiology, The First Affiliated Hospital of DalianMedical University, Dalian, China (Y.C., H.L., D.D., Y.Q., R.P., A.L., B.S.).
Academic Radiology
|April 25, 2025
Summary
A new radiomic model using MRI T1CE imaging effectively differentiates glioblastoma (GBM) from solitary brain metastasis (SBM). This model, focusing on the brain-tumor interface, improves diagnostic accuracy over human experts.
Area of Science:
- Radiology
- Oncology
- Medical Imaging Analysis
Background:
- Glioblastoma (GBM) and solitary brain metastasis (SBM) share similar MRI radiomics features.
- Accurate differentiation is critical due to differing treatments and prognoses.
- Current diagnostic methods require improvement for precise tumor classification.
Purpose of the Study:
- To develop and validate a diagnostic model for differentiating GBM from SBM.
- To utilize radiomic features from the T1-weighted contrast-enhanced (T1CE) sequence.
- To focus on the 10 mm brain-tumor interface (BTI) region for enhanced discrimination.
Main Methods:
- Retrospective analysis of 226 GBM and 206 SBM T1CE MRI scans from three centers.
- Training and testing datasets established; 10 mm BTI ROIs extracted.
- Radiomic features extracted, selected, and modeled using logistic regression; SHAP for visualization.
- Model performance compared against three experienced radiologists using the DeLong test.
Main Results:
- Ten radiomic features were selected for the final model.
- The logistic regression model achieved an AUC of 0.893 (training) and 0.808 (test).
- The radiomic model outperformed two of three radiologists (AUCs 0.699, 0.740, 0.789) and showed significant improvement over the least experienced radiologist.
Conclusions:
- The radiomic model based on the 10 mm BTI effectively differentiates GBM and SBM.
- This approach captures tumor heterogeneity, enhancing diagnostic performance.
- The model serves as a valuable tool to assist clinicians in differentiating these brain tumors.
