Related Experiment Video
Updated: Aug 8, 2026
06:35
Basics of Multivariate Analysis in Neuroimaging Data
Published on: July 24, 2010
Analysis of MTR histograms in multiple sclerosis using principal components and multiple discriminant analysis
J Dehmeshki1, A C Ruto, S Arridge
1NMR Research Unit, Department of Clinical Neurology, Institute of Neurology, University College London, London, UK. j.dehmeshki@ion.ucl.ac.uk
Magnetic Resonance in Medicine
|September 11, 2001
Summary
New analysis of Magnetization Transfer Ratio (MTR) histograms improves characterization of white matter disease like multiple sclerosis (MS). Advanced methods like Principal Component Analysis (PCA) and Multiple Discriminant Analysis (MDA) offer superior classification and correlation with disease severity.
Area of Science:
- Neuroimaging
- Biophysics
- Medical Statistics
Background:
- Magnetization transfer ratio (MTR) histograms are valuable for detecting diffuse white matter changes in neurological diseases.
- Existing MTR histogram analysis methods have limitations in correlating with clinical disability, particularly in multiple sclerosis (MS).
Purpose of the Study:
- To develop and evaluate a novel MTR histogram analysis method for improved characterization of white matter disease.
- To assess the correlation of the new method with clinical disability measured by the Expanded Disability Status Scale (EDSS).
- To compare the efficacy of Principal Component Analysis (PCA) and Multiple Discriminant Analysis (MDA) against traditional MTR histogram features for classifying MS subgroups.
Main Methods:
- Utilized MTR histogram data from a central 60-mm slab of brain tissue.
- Applied Principal Component Analysis (PCA) and Multiple Discriminant Analysis (MDA) for feature extraction and classification.
- Compared PCA and MDA performance against traditional histogram features (peak height, peak location).
- Performed multiple linear regression analysis of principal components (PCs) against EDSS scores.
Main Results:
- The novel method demonstrated improved correlation with the Expanded Disability Status Scale (EDSS).
- PCA and MDA yielded superior classification results compared to traditional MTR histogram features.
- Binary classification success rates between control and MS subgroups using MDA ranged from 75-95%.
- Multiple regression analysis showed significant correlations between EDSS and the first three PCs (r=0.83 for SPMS, r=0.80 for all MS patients).
Conclusions:
- Advanced analytical techniques, specifically PCA and MDA, significantly enhance the diagnostic and prognostic capabilities of MTR histogram analysis in white matter diseases like MS.
- The developed MR-based measure shows strong potential for objectively assessing disease severity and progression in MS patients.
- This approach offers a more refined tool for differentiating between normal controls and various MS subgroups.
