Hybrid Deep Learning Framework for Alzheimer's Detection: Perona-Malik Filter-Based FAWT Feature Extraction with ResNet-50 on MRI Images
Abstract
Alzheimer's disease (AD), the leading cause of dementia, burdens global healthcare. Early MRI-based detection of brain changes is vital for intervention. This study introduces PMF-FAWT for robust AD diagnosis. MRI images are denoised using Perona-Malik filters, decomposed using FAWT for multiscale features, and deep features are extracted with ResNet-50. These features are then refined through statistical ranking and PCA. LS-SVM with an RBF kernel is used for classification and evaluation on public datasets via 10-fold cross-validation. Achieves 98.6% accuracy, 98.38% sensitivity, and 98.0% specificity, outperforming traditional wavelets. PMF-FAWT offers an efficient, interpretable tool for accurate AD detection and management.
Fields of Study & Subjects
Brain Tumor Detection and Classification
Advanced Neural Network Applications
Retinal Imaging and Analysis
Publication details
- Journal / source
- IETE Technical Review
- Volume
- 43
- Issue
- 2
- Pages
- 177–188
- Publication date
- 5 January 2026
- Language
- EN