Development of Convolutional Neural Network Model for Classification of Alzheimer's Disease Using Inception V3 Architecture

Authors

  • Samuel Yinka Olatunde Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria https://orcid.org/0009-0002-5846-787X
  • Olatayo Moses Olaniyan Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria https://orcid.org/0000-0002-7349-7500
  • Micheal Abimbola Oladosu Department of Biochemistry Faculty of Basic Medical Sciences University of Lagos, Lagos, Nigeria , Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria https://orcid.org/0009-0000-5098-1247
  • Olaoluwa John Adeleke Department of Electrical Engineering, The Polytechnic Ibadan, Ibadan, Oyo State, Nigeria. https://orcid.org/0009-0009-6372-4640
  • Emmanuel Nkansah Department of Accounting, Economics and Finance, School of Business, La Sierra University, Riverside, United States.
  • Moses Adondua Abah Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria
  • Abiodun Rasak Folorunsho 1Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria https://orcid.org/0009-0002-7012-9486
  • Stephen Olaoluwapo Adeleye Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria https://orcid.org/0009-0002-5456-1949
  • Olalekan Joseph Dada Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria
  • Shola David Omoseeye Department of Anatomy, Faculty of Basic Medical Sciences, College of Medicine, Ekiti State University, Ado Ekiti, Ekiti State, Nigeria. https://orcid.org/0000-0003-3305-3785
  • Angel Ojimaojo Ekele Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria

DOI:

https://doi.org/10.12856/JHIA-2026-v13-i1-610

Abstract

Background: Alzheimer's disease (AD) presents significant diagnostic challenges due to its complex and varied symptoms. Traditional diagnostic methods, often involving subjective assessments and imaging techniques, can delay diagnosis and treatment. Convolutional Neural Networks (CNNs) offer promising solutions by automating detailed feature extraction from medical images. This study aimed to develop a robust CNN model using the Inception V3 architecture for accurate Alzheimer's disease classification and severity staging from medical imaging data.

Methods: A comprehensive dataset of 6,400 brain MRI images was obtained from Kaggle, categorised into four severity classes: Non-Demented (n=3,200), Very Mild Demented (n=2,240), Mild Demented (n=896), and Moderate Demented (n=64). The dataset was split into training (70%, n=4,480), validation (15%, n=960), and testing (15%, n=960) sets. The Inception V3 model was implemented using transfer learning techniques in Google Colaboratory with GPU acceleration. Training was conducted over 100 epochs with a batch size of 30, using the Adam optimiser and the categorical cross-entropy loss function. Data preprocessing included image resizing, normalisation, and augmentation techniques to enhance generalizability.

Results: The Inception V3 model demonstrated exceptional performance metrics after 100 epochs of training. The model achieved a validation accuracy of 98.52%, precision of 98.66%, recall of 98.21%, and F1-score of 98.35%. Training accuracy progressed from 67.23% initially to nearly 100% by the final epoch, while validation loss decreased progressively throughout training. The model showed superior performance compared to existing approaches, outperforming the SE-Net Model using SVM Classifier (98.37% accuracy) reported in recent literature.

Conclusions: This study successfully developed and validated a CNN model using Inception V3 architecture for Alzheimer's disease classification and severity staging. The model's high-performance metrics demonstrate its potential as a reliable tool for early diagnosis and personalised treatment planning. The effective utilization of Inception V3's advanced architecture demonstrates how deep learning techniques can transform diagnostic procedures and improve the early detection of Alzheimer's disease, ultimately enhancing patient care outcomes.

Downloads

Download data is not yet available.

Author Biographies

  • Samuel Yinka Olatunde , Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

    Graduate Student

    Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

  • Olatayo Moses Olaniyan , Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

    Senior Lecturer

    Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

  • Micheal Abimbola Oladosu, Department of Biochemistry Faculty of Basic Medical Sciences University of Lagos, Lagos, Nigeria , Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria

    Acting  Director

    Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria

  • Olaoluwa John Adeleke, Department of Electrical Engineering, The Polytechnic Ibadan, Ibadan, Oyo State, Nigeria.

    Graduate Student

    Department of Electrical Engineering, The Polytechnic Ibadan, Ibadan, Oyo State, Nigeria.

  • Emmanuel Nkansah, Department of Accounting, Economics and Finance, School of Business, La Sierra University, Riverside, United States.

    Graduate Student

    Department of Accounting, Economics and Finance, School of Business, La Sierra University, Riverside, United States.

  • Moses Adondua Abah, Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria

    Technical Head

    Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria

  • Abiodun Rasak Folorunsho, 1Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

    Graduate Student

    Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

  • Stephen Olaoluwapo Adeleye, Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

    Graduate Student

    Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

  • Olalekan Joseph Dada, Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

    Graduate Student

    Department of Computer Engineering, Faculty of Engineering, Federal University Oye-Ekiti, Oye-Ekiti, Ekiti State, Nigeria

  • Shola David Omoseeye, Department of Anatomy, Faculty of Basic Medical Sciences, College of Medicine, Ekiti State University, Ado Ekiti, Ekiti State, Nigeria.

    Graduate Student

    Department of Anatomy, Faculty of Basic Medical Sciences, College of Medicine, Ekiti State University, Ado Ekiti, Ekiti State, Nigeria.

     

  • Angel Ojimaojo Ekele, Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria

    Research Assistant

    Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos State, Nigeria

Downloads

Published

2026-10-05

Issue

Vol. 13 No. 2 (2026)

Section

Research Article

How to Cite

[1]
Olatunde , S.Y. et al. 2026. Development of Convolutional Neural Network Model for Classification of Alzheimer’s Disease Using Inception V3 Architecture. Journal of Health Informatics in Africa. 13, 2 (Oct. 2026), 25–40. DOI:https://doi.org/10.12856/JHIA-2026-v13-i1-610.

Most read articles by the same author(s)