Related Experiment Video
Updated: Jul 1, 2026
06:48
On-Site Sampling and Extraction of Brain Tumors for Metabolomics and Lipidomics Analysis
Published on: May 31, 2020
Prediction of Molecular Mutations in Diffuse Low-Grade Gliomas using MR Imaging Features
Zeina A Shboul1, James Chen2,3, Khan M Iftekharuddin4
1Vision Lab, Electrical & Computer Engineering, Old Dominion University, Norfolk, VA, USA.
Scientific Reports
|March 1, 2020
Summary
This study introduces non-invasive imaging techniques to predict molecular mutations in low-grade gliomas (LGG), offering an alternative to invasive biopsies. The developed models show promising accuracy in identifying key genetic alterations, aiding in LGG classification.
Area of Science:
- Neuro-oncology
- Medical imaging
- Computational biology
Background:
- Low-grade gliomas (LGG) reclassification relies on molecular mutations, typically requiring invasive biopsies.
- Biopsies carry risks of sampling error and may not represent the entire tumor heterogeneity.
- Non-invasive imaging offers a comprehensive approach to assess tumor characteristics.
Purpose of the Study:
- To develop non-invasive imaging-based prediction models for key molecular mutations in LGG.
- To correlate magnetic resonance imaging (MRI) features with specific molecular alterations (MGMT methylation, IDH mutation, 1p/19q co-deletion, ATRX mutation, TERT mutations).
- To evaluate the potential of imaging features as predictive biomarkers for LGG molecular classification.
Main Methods:
- Extraction of imaging features from MRI data, including texture, fractal, multi-resolution fractal, and volumetric features.
- Development of prediction models using nested leave-one-out cross-validation for feature selection and performance estimation.
- Training and validation of models to predict MGMT methylation, IDH mutations, 1p/19q co-deletion, ATRX mutations, and TERT mutations.
Main Results:
- Prediction models achieved significant test performance: AUC of 0.83±0.04 (MGMT methylation), 0.84±0.03 (IDH mutations), 0.80±0.04 (1p/19q co-deletion), 0.70±0.09 (ATRX mutation), and 0.82±0.04 (TERT mutations).
- Fractal imaging features demonstrated a significant impact on the predictive performance for MGMT methylation, IDH mutations, 1p/19q co-deletion, and ATRX mutations.
- The study successfully correlated computed imaging features with LGG molecular mutation types.
Conclusions:
- Non-invasive imaging analysis shows potential for predicting molecular subtypes of LGG.
- Imaging features, particularly fractal characteristics, can serve as valuable biomarkers for LGG molecular classification.
- This approach may reduce the need for invasive tumor sampling in LGG diagnosis and management.
