Related Experiment Video
Updated: Jan 3, 2026
13:12
Translational Brain Mapping at the University of Rochester Medical Center: Preserving the Mind Through Personalized Brain Mapping
Published on: August 12, 2019
46.2K
Pre-surgical connectome features predict IDH status in diffuse gliomas.
Shelli R Kesler1,2,3, Rebecca A Harrison4,3, Melissa L Petersen4
1Cancer Neuroscience Laboratory, School of Nursing, The University of Texas at Austin, Austin, Texas, USA.
Oncotarget
|November 20, 2019
Summary
Connectomics, a brain network analysis, can non-invasively predict isocitrate dehydrogenase (IDH) gene status in glioma patients using MRI. Random forest machine learning models demonstrated high accuracy, aiding surgical planning and patient counseling.
Area of Science:
- Neuroimaging
- Oncology
- Machine Learning
Background:
- Gliomas are common malignant brain tumors.
- Tumor molecular characteristics, like IDH gene mutations, impact patient outcomes.
- Preoperative IDH genotype determination is crucial for personalized treatment and clinical trials.
Purpose of the Study:
- To evaluate connectomics for non-invasive IDH genotype prediction in gliomas using MRI.
- To assess the performance of machine learning models in predicting IDH status.
Main Methods:
- Retrospective analysis of 234 adult patients' T1-weighted MRI data.
- Extraction of 93 whole-brain connectome features.
- Evaluation of four machine learning models for IDH genotype prediction.
Main Results:
- Connectomics achieved high predictive performance (AUC 0.76-0.94).
- Random forest model significantly outperformed other algorithms (p < 0.01).
- Feature selection and inclusion of age/location did not alter random forest performance.
Conclusions:
- Connectomics is a feasible method for preoperative IDH genotype prediction in gliomas.
- Random forest is an effective machine learning approach for this task.
- Connectomics offers insights into genotype effects on brain network organization.
