AD2C-SG-TL: Alzheimer's Disease Detection and Classification Based on Stacked Generalization and Transfer Learning Approaches

Authors

Mariam Gamal Alboghdady, Computers and Control Systems Engineering Department, Faculty of Engineering, Mansoura University, Mansoura 35516, Egypt
Amira Y. Haikal, Computers and Control Systems Engineering Department, Faculty of Engineering, Mansoura University, Mansoura 35516, Egypt
Hesham H. Gad, Computers and Control Systems Engineering Department, Faculty of Engineering, Mansoura University, Mansoura 35516, Egypt
Noha A. Sakr, Computers and Control Systems Engineering Department, Faculty of Engineering, Mansoura University, Mansoura 35516, EgyptFollow

Subject Area

Computer Science

Article Type

Original Study

Abstract

Alzheimer's disease (AD) is a degenerative neurologic illness that causes brain atrophy and cell death. Although there is no cure for AD, diagnosing its onset can be very beneficial in the medical field.This paper presents a deep ensemble learning framework for classifying AD stages. Transfer learning (TL) is applied using eight pretrained convolutional neural networks (CNNs) (i.e., VGG16, VGG19, ResNet50V2, MobileNet, DenseNet121, DenseNet169, Xception and MobileNetV2). Stackedgeneralization ensembles techniques are used to provide greater generalization by combining finetuned models with five ensemble models. Using five different stacked ensembles (SE) models to improve the generalization.The ensemble model created by combining all the finetuned networks obtained a state-of-the-art AD accuracy detection score of 99.92%. The specificity and sensitivity rates are 99.94% and 99.92%, respectively, highlighting the robustness of stacked ensembles. Moreover, the winning proposed ensembled model is compared against seven state of the art techniques to outperform them with 99.92% accuracy. Two other proposed models win the competition against another ensemble learning 3D state of the art model with an accuracy of 99.83% ,99.75%.

Keywords

Alzheimer's disease, Stacked generalization, Deep learning, Transfer learning, ADNI

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 License.

Recommended Citation

Alboghdady, Mariam Gamal; Haikal, Amira Y.; Gad, Hesham H.; and Sakr, Noha A. (2025) "AD2C-SG-TL: Alzheimer's Disease Detection and Classification Based on Stacked Generalization and Transfer Learning Approaches," Mansoura Engineering Journal: Vol. 50 : Iss. 4 , Article 6.
Available at: https://doi.org/10.58491/2735-4202.3277