Related Experiment Video
Updated: Jan 6, 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
Probabilistic template matching for detecting resting-state functional MRI language network in brain tumor patients
Jian Ming Teo1,2, Vinodh A Kumar3, Alexander M Khalaf3
1Department of Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA.
Medical Physics
|November 13, 2025
Summary
This study developed probabilistic language templates to improve the detection of language networks in brain tumor patients using resting-state fMRI. These new methods enhance accuracy in identifying language networks compared to traditional approaches.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Brain Network Mapping
Background:
- Detecting language networks with resting-state fMRI (rs-fMRI) using independent component analysis (ICA) is challenging in brain tumor patients due to intersubject variations.
- Standard template matching methods struggle to account for individual differences in brain anatomy and tumor-related changes.
Purpose of the Study:
- To develop and validate methods for incorporating intersubject variation into language network detection in brain tumor patients.
- To utilize a probabilistic language atlas for improved template matching in functional MRI (fMRI) analysis.
Main Methods:
- Retrospective analysis of 79 brain tumor patients undergoing task-based (tb)-fMRI and rs-fMRI.
- Development of binary and probabilistic language templates from an atlas at varying thresholds (τ).
- Comparison of template matching methods including goodness-of-fit (GOF), weighted GOF (wGOF), and Jensen-Shannon distance (JSD), using Dice coefficient and Pearson correlation against tb-fMRI.
Main Results:
- Probabilistic templates significantly improved language network detection (higher Dice coefficients and Pearson correlations) compared to a control binary template across τ = 0% to 35%.
- Peak average Dice coefficients reached 0.349 (wGOF) and 0.350 (JSD), substantially higher than the control's 0.247.
- Qualitative assessment showed significantly superior performance (p < 0.05) with probabilistic templates, increasing correct language network identification from 58-52 to 69-73 cases.
Conclusions:
- Probabilistic templates derived from a language tb-fMRI atlas effectively address intersubject variation in brain tumor patients.
- The proposed probabilistic template matching methods enhance the detection accuracy of rs-fMRI ICA language networks in this population.
- This approach offers a more reliable tool for presurgical language mapping in neuro-oncology.
