A Deep Learning-Based Ensemble Method for Early Diagnosis of Alzheimer’s Disease using MRI Images
Problem
Early diagnosis of Alzheimer's disease is difficult, and its growing prevalence imposes major costs on individuals and society.
Why it matters
Earlier, accurate AD diagnosis could reduce the growing personal and societal burden of the disease.
What the AI did
An ensemble of the six best CNN-based classifiers was trained on MRI images and evaluated on ADNI data, then validated on a local dataset.
Results
Accuracies of 98.57 (NC/AD), 96.37 (NC/EMCI), 94.22 (EMCI/LMCI), 99.83 (LMCI/AD), 93.88 (four-way) and 93.92 (three-way); validated on a local dataset.
Impact
Demonstrated high-accuracy MRI-based AD classification on ADNI and an independent local dataset.
Record
| Venue | Neuroinformatics |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | Sina Fathi, Ali Ahmadi, Afsaneh Dehnad, Mostafa Almasi‐Dooghaee, Melika Sadegh, for the Alzheimer’s Disease Neuroimaging Initiative |
| Citations | 95 (OpenAlex, at import time) |
| References | 55 |
| Model or method | Ensemble of six CNN-based classifiers |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1007/s12021-023-09646-2 |
| Open access | hybrid |
| Keywords | Alzheimer's disease, deep learning, ensemble, MRI, CNN, early diagnosis |
Importance rating
significant — thousands to millions affected
Abstract cites growing prevalence of Alzheimer's disease and resulting costs to individuals and society; early diagnosis addresses a major disease burden.