Facebook pixel tracking

The Science and Information (SAI) Organization publishes open-access peer-reviewed journals in computer science and artificial intelligence.

Contact Info
Website thesai.org
Follow Us
Contact Info
Follow Us
Research Article | Open Access |

Investigating the Effect of the Data Partitioning Protocol on Automatic Alzheimer’s Disease Detection from Brain MRI Using Deep Learning

Author 1: Fatima Zahra Salmam Author 2: Sanae Belfrouh Author 3: Lamya Ouahmidi Author 4: Rahhal Errattahi
International Journal of Advanced Computer Science and Applications (IJACSA) · Vol. 17, No. 10 · Published 2026

DOI: 10.14569/IJACSA.2026.01710118

Abstract

Computer vision and deep learning have been widely used to automatically detect and analyze many diseases from medical images. Some of these include tumor detection, diabetic retinopathy classification, nodules classification, and the screening of thoracic pathologies. In this work, we are investigating the application of deep learning techniques to the detection of Alzheimer’s disease from brain MRI images. The general purpose was to offer an aided diagnosis system able to distinguish the three diagnostic groups of the ADNI database, namely Alzheimer’s disease, cognitively normal subjects and subjects with mild cognitive impairment. An Inception-V3 model pretrained on ImageNet was adapted to the task and fine-tuned by gradually unfreezing its top convolutional blocks. To address the problem of the imbalance between the diagnostic classes, and given the limited size of the available cohort, we compared three strategies: class weights, data augmentation, and a subject-level version of SMOTE. Then, given the nature of the task, we opted for a double evaluation of the developed models, first with a partition performed at the subject level and then with a partition performed at the slice level. On a cohort of 106 ADNI subjects, the best three-class accuracy under a strict subject-level partition was 67.77% (SMOTE), whereas the same models reached more than 90% with a slice-level partition. The difference between imbalance strategies (up to 13 points) vanished under leakage (less than 0.5 point), showing that, in this setting, the partitioning protocol had a larger influence on the reported accuracy than the strategy used.

Keywords

How to Cite this Article

Salmam, F. Z., Belfrouh, S., Ouahmidi, L., & Errattahi, R. (2026). Investigating the Effect of the Data Partitioning Protocol on Automatic Alzheimer’s Disease Detection from Brain MRI Using Deep Learning. International Journal of Advanced Computer Science and Applications, 17(10). https://doi.org/10.14569/IJACSA.2026.01710118

Salmam, Fatima Zahra, et al.. "Investigating the Effect of the Data Partitioning Protocol on Automatic Alzheimer’s Disease Detection from Brain MRI Using Deep Learning." International Journal of Advanced Computer Science and Applications, vol. 17, no. 10, 2026, https://doi.org/10.14569/IJACSA.2026.01710118.

@article{Salmam2026,
  title     = {Investigating the Effect of the Data Partitioning Protocol on Automatic Alzheimer’s Disease Detection from Brain MRI Using Deep Learning},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {17},
  number    = {10},
  year      = {2026},
  publisher = {The Science and Information Organization},
  author    = {Fatima Zahra Salmam and Sanae Belfrouh and Lamya Ouahmidi and Rahhal Errattahi},
  doi       = {10.14569/IJACSA.2026.01710118},
  url       = {https://doi.org/10.14569/IJACSA.2026.01710118}
}

Open Access — licensed under a Creative Commons Attribution 4.0 International License. Unrestricted use, distribution, and reproduction in any medium, even commercially, as long as the original work is properly cited.

Call for Papers - Important Dates

Volume 17 No. 12, December 2026
Paper Submission Due November 25, 2026
Review Notification December 15, 2026
Publication Date December 31, 2026