Related Experiment Video
Updated: Apr 2, 2026
09:33
Diffusion Tensor Magnetic Resonance Imaging in the Analysis of Neurodegenerative Diseases
Published on: July 28, 2013
29.5K
FEATURE SELECTION IMPROVES THE ACCURACY OF CLASSIFYING ALZHEIMER DISEASE USING DIFFUSION TENSOR IMAGES.
Ayşe Demirhan1, Talia M Nir2, Artemis Zavaliangos-Petropulu2
1Electronics & Computer Technology, Faculty of Technology, Gazi University, Ankara, Turkey ; Imaging Genetics Center, Keck School of Medicine of USC, Marina del Rey, CA, USA.
Proceedings. IEEE International Symposium on Biomedical Imaging
|September 29, 2015
Summary
Diffusion tensor imaging (DTI) reveals white matter (WM) changes in Alzheimer's disease (AD). This study used DTI fractional anisotropy (FA) to accurately classify AD patients, mild cognitive impairment, and healthy controls.
Area of Science:
- Neuroimaging
- Neurology
- Medical Image Analysis
Background:
- Alzheimer's disease (AD) is associated with widespread white matter (WM) abnormalities.
- Standard MRI may not detect subtle WM changes characteristic of AD.
- Diffusion tensor imaging (DTI) offers a method to investigate these WM abnormalities.
Purpose of the Study:
- To classify Alzheimer's disease (AD) patients, mild cognitive impairment (MCI) individuals, and healthy elderly controls using DTI.
- To identify specific white matter (WM) pathways affected in AD.
- To enhance classification accuracy by selecting discriminative WM voxels.
Main Methods:
- Analysis of image-wide DTI fractional anisotropy (FA) maps.
- Classification using Support Vector Machine (SVM) driven by voxelwise FA and WM region of interest (ROI) averages.
- Feature selection using the ReliefF algorithm to identify discriminative WM voxels.
Main Results:
- Improved classification accuracy by up to 15% for AD, MCI, and control groups.
- Identification of specific WM clusters affected in AD using the ReliefF algorithm.
- Highlighting WM pathways impacted by AD that may be missed by traditional ROI analysis.
Conclusions:
- DTI-based analysis, particularly with voxel selection, can effectively differentiate between AD, MCI, and healthy controls.
- The ReliefF algorithm can identify specific WM tracts crucial for accurate AD classification.
- This approach provides a more comprehensive understanding of WM pathology in Alzheimer's disease.
