Multi-class Alzheimer’s disease classification using VGG19 transfer learning with hybrid Harris Hawks Optimization and random forest
Abstract
Accurate cognition-state classification of Alzheimer’s disease (AD), Mild Cognitive Impairment (MCI), and Cognitively Normal (CN) is essential for early evaluation of the disease. Existing deep-learning approaches often retain redundant features, depend on dataset-specific configurations, and provide limited methodological optimization. This study proposes a transfer-learning-based MRI classification framework integrating spatial and intensity normalization, histogram equalization, double bilateral filtering, VGG19-based deep feature extraction, Harris Hawks Optimization with Simulated Annealing (HHO-SA) for feature selection, and Random Forest (RF) classification. The framework was implemented using Python and PyTorch and evaluated on the ADNI1 Complete 1Yr 1.5T dataset, containing 2,294 MRI observations from 639 subjects. Subject-level partitioning was employed to minimize data leakage. The proposed framework achieved 98.70% accuracy, 98.55% macro-precision, 98.93% macro-recall, 99.28% macro-specificity, 98.73% macro-F1, and 99.40% macro-ROC-AUC. Compared with the strongest reported benchmark accuracy of 98.20%, this study developed into 0.50%. The findings demonstrate effective retrospective three-class MRI classification and provide a reproducible computational framework for AD research.
Related
- Multimodal patch-based deep learning for early Alzheimer’s disease classification using MRI and PET
- An uncertainty-aware multimodal deep learning framework for Alzheimer’s disease classification using MRI and electronic health records
- Hybrid deep learning and meta learning for Alzheimer disease classification via multimodal radiomic and visual features
- Montage-agnostic federated learning for privacy-preserving Alzheimer’s disease classification from heterogeneous multi-site EEG
- Machine learning classification and regional differentiation of neuropathologically-confirmed Alzheimer’s disease and comorbid Lewy body disease
BibTeX
@article{selvi2026multiclass,
title = {{Multi-class Alzheimer’s disease classification using VGG19 transfer learning with hybrid Harris Hawks Optimization and random forest}},
author = {G. Anbu Selvi and T. Karthick},
journal = {Scientific Reports},
year = {2026},
month = oct,
doi = {10.1038/s41598-026-74521-4},
url = {https://www.nature.com/articles/s41598-026-74521-4},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Multi-class Alzheimer’s disease classification using VGG19 transfer learning with hybrid Harris Hawks Optimization and random forest
AU - G. Anbu Selvi
AU - T. Karthick
PY - 2026
DA - 2026/10/16
JO - Scientific Reports
DO - 10.1038/s41598-026-74521-4
UR - https://www.nature.com/articles/s41598-026-74521-4
AB - Accurate cognition-state classification of Alzheimer’s disease (AD), Mild Cognitive Impairment (MCI), and Cognitively Normal (CN) is essential for early evaluation of the disease. Existing deep-learning approaches often retain redundant features, depend on dataset-specific configurations, and provide limited methodological optimization. This study proposes a transfer-learning-based MRI classification framework integrating spatial and intensity normalization, histogram equalization, double bilateral filtering, VGG19-based deep feature extraction, Harris Hawks Optimization with Simulated Annealing (HHO-SA) for feature selection, and Random Forest (RF) classification. The framework was implemented using Python and PyTorch and evaluated on the ADNI1 Complete 1Yr 1.5T dataset, containing 2,294 MRI observations from 639 subjects. Subject-level partitioning was employed to minimize data leakage. The proposed framework achieved 98.70% accuracy, 98.55% macro-precision, 98.93% macro-recall, 99.28% macro-specificity, 98.73% macro-F1, and 99.40% macro-ROC-AUC. Compared with the strongest reported benchmark accuracy of 98.20%, this study developed into 0.50%. The findings demonstrate effective retrospective three-class MRI classification and provide a reproducible computational framework for AD research.
SN - 2045-2322
ER -