Related Experiment Video
Updated: Nov 19, 2025
12:24
Computed Tomography-guided Time-domain Diffuse Fluorescence Tomography in Small Animals for Localization of Cancer Biomarkers
Published on: July 17, 2012
12.7K
Machine learning-based differentiation between multiple sclerosis and glioma WHO II°-IV° using O-(2-[18F]
Sied Kebir1,2,3, Laurèl Rauschenbach3,4, Manuel Weber5
1Division of Clinical Neurooncology, Department of Neurology, University Hospital Essen, University Duisburg-Essen, Hufelandstrasse 55, 45147, Essen, Germany.
Journal of Neuro-Oncology
|January 27, 2021
Summary
This study shows that FET-PET imaging combined with machine learning can accurately differentiate multiple sclerosis (MS) from glioma. The machine learning approach significantly improved diagnostic performance compared to standard analysis.
Area of Science:
- Neurology
- Radiology
- Oncology
Background:
- Differentiating multiple sclerosis (MS) from glioma is diagnostically challenging.
- Accurate differentiation is crucial for appropriate patient management and treatment.
Purpose of the Study:
- To evaluate the diagnostic significance of FET-PET imaging combined with machine learning for distinguishing MS from glioma II°-IV°.
- To assess the performance enhancement offered by machine learning algorithms in this differentiation.
Main Methods:
- Retrospective analysis of patients undergoing FET-PET for suspected glioma.
- Inclusion of histologically confirmed glioma II°-IV° and MS cases.
- Determination of tumor-to-brain ratio (TBR) features and application of a support vector machine (SVM) algorithm.
- Receiver Operating Characteristic (ROC) analysis to evaluate model performance.
Main Results:
- A total of 41 patients (34 glioma, 7 MS) were analyzed.
- Significantly higher TBR values were observed in the glioma group compared to MS.
- SVM-based machine learning improved the differentiation accuracy from an AUC of 0.79 to 0.94.
Conclusions:
- FET-PET imaging shows diagnostic potential in differentiating MS from glioma II°-IV°.
- Machine learning approaches, specifically SVM, significantly enhance the classification performance.
