Related Experiment Video
Updated: Jun 23, 2026
14:27
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
Linking functional and structural brain images with multivariate network analyses: a novel application of the partial
Kewei Chen1, Eric M Reiman, Zhongdan Huan
1Banner Alzheimer's Institute and the Banner Good Samaritan PET Center, Phoenix, AZ 85006, USA. kchen@math.la.asu.edu
Neuroimage
|April 28, 2009
Summary
A new multimodal network analysis method effectively distinguishes older from younger adults using brain imaging data. This powerful technique integrates complementary imaging patterns for enhanced individual brain analysis.
Area of Science:
- Neuroimaging
- Biostatistics
- Network Analysis
Background:
- Distinguishing age-related brain patterns is crucial for understanding cognitive aging.
- Existing methods for analyzing multimodal brain imaging data have limitations in power and sensitivity.
Purpose of the Study:
- To introduce a novel multimodal multivariate network analysis for characterizing brain image patterns.
- To demonstrate the method's superior ability to differentiate age groups compared to established techniques.
Main Methods:
- Utilizes partial least square (PLS) algorithm on complementary co-registered brain images (PET and MRI).
- Calculates a combined latent variable maximizing covariance across all variables.
- Employs a computationally feasible approach for singular value decomposition of high-dimensional covariance matrices.
Main Results:
- Successfully distinguished older from younger adults with high statistical significance (p=4e-12) and no overlap.
- Demonstrated greater power than univariate SPM, multimodal SPM, and single-modality PLS.
- Eliminated the need for multiple comparisons correction.
Conclusions:
- The proposed multimodal network analysis offers a powerful new approach for brain image analysis.
- This technique can integrate diverse complex datasets beyond brain imaging.
- It enables more robust characterization of individual brain patterns related to state or traits.
