Related Experiment Video
Updated: May 1, 2026
09:33
Diffusion Tensor Magnetic Resonance Imaging in the Analysis of Neurodegenerative Diseases
Published on: July 28, 2013
28.6K
DUAL MULTI-ATLAS REPRESENTATION ALIGNMENT FOR BRAIN DISORDER DIAGNOSIS USING MORPHOLOGICAL CONNECTOME.
Kangfu Han1, Dan Hu1, Jiale Cheng1
1Department of Radiology and Biomedical Research Imaging Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA.
Summary
This study introduces a dual multi-atlas representation alignment approach (DMAA) to improve brain disorder diagnosis by integrating diverse brain atlas data. DMAA effectively reduces variability and harmonizes features for more accurate diagnostic insights.
Area of Science:
- Neuroimaging
- Computational Neuroscience
- Medical Informatics
Background:
- Morphological connectome analysis in structural MRI is crucial for diagnosing brain disorders.
- Integrating data from multiple brain atlases presents challenges due to variations in region definitions and inter-atlas variability.
- Existing methods struggle to effectively fuse complementary information from diverse atlases while mitigating bias.
Purpose of the Study:
- To propose a novel dual multi-atlas representation alignment approach (DMAA) for enhanced brain disorder diagnosis.
- To effectively integrate complementary information from multiple brain atlases.
- To reduce inter-atlas variability and manage anatomical differences for improved diagnostic accuracy.
Main Methods:
- Developed a dual multi-atlas representation alignment approach (DMAA).
- Minimized maximum mean discrepancy to align multi-atlas representations into a unified distribution.
- Applied optimal transport to harmonize region-wise differences and preserve cross-atlas relationships.
Main Results:
- Demonstrated the effectiveness of DMAA on multiple large-scale datasets (ADNI, PPMI, ADHD200, SchizConnect).
- Successfully reduced inter-atlas variability and enhanced feature fusion.
- Improved diagnostic accuracy for brain disorders using multi-atlas morphological connectome.
Conclusions:
- The proposed DMAA method offers a robust solution for integrating multi-atlas data in brain disorder diagnosis.
- DMAA effectively addresses challenges related to atlas variability and feature fusion.
- This approach holds significant potential for advancing neuroimaging-based diagnostics for various brain conditions.
