Related Experiment Video
Updated: Jun 6, 2026
05:17
Hybrid PET/MRI Imaging of Alzheimer's Disease Based on 18F-AV-1451
Published on: April 18, 2025
Feature selection using factor analysis for Alzheimer's diagnosis using 18F-FDG PET images.
D Salas-Gonzalez1, J M Górriz, J Ramírez
1Department of Signal Theory, Networking and Communications, ETSIIT 18071, University of Granada, Granada, Spain. dsalas@ugr.es
Medical Physics
|December 17, 2010
Summary
This study introduces a computer-aided diagnosis method for early Alzheimer's disease (AD) detection. The technique achieves high accuracy, outperforming other methods in classifying normal controls, mild cognitive impairment, and AD subjects.
Area of Science:
- Medical Imaging
- Neurology
- Computer-Aided Diagnosis
Background:
- Alzheimer's disease (AD) diagnosis relies on accurate identification of neurodegenerative changes.
- Early detection of AD is crucial for timely intervention and management.
- 18F-FDG PET imaging offers insights into brain metabolism, valuable for AD assessment.
Purpose of the Study:
- To develop and evaluate a computer-aided diagnosis (CAD) technique for enhancing early Alzheimer's disease (AD) detection accuracy.
- To analyze 18F-FDG PET images from normal controls (NC), mild cognitive impairment (MCI), and AD subjects.
- To compare the proposed CAD technique against existing methods for classification performance.
Main Methods:
- Voxel selection using t-tests and feature dimension reduction via factor analysis.
- Utilizing factor loadings as features for classification.
- Employing three classifiers: Gaussian mixture models (linear and quadratic discriminant functions) and a support vector machine (SVM) with a linear kernel.
Main Results:
- Achieved up to 95% accuracy in classifying normal controls (NC) versus Alzheimer's disease (AD) subjects.
- Obtained 88% accuracy for NC-MCI classification and 86% for NC-MCI-AD classification using SVM with a linear kernel.
- Demonstrated superior classification performance compared to voxel-as-features and PCA-based approaches.
Conclusions:
- The proposed computer-aided diagnosis methodology shows significant potential for accurate early Alzheimer's disease detection.
- The factor analysis-based feature extraction combined with SVM offers improved classification accuracy.
- This technique provides a promising tool for differentiating between normal cognition, mild cognitive impairment, and Alzheimer's disease stages.