Deep Learning Approaches for Early Detection and Classification of Alzheimer's Disease

Diqing Xu

2024

Abstract

Alzheimer's Disease (AD) is a neurodegenerative condition that presents major obstacles to early diagnosis and classification. This study proposes a new deep learning-based method to classify preprocessed brain MRI scans, incorporating techniques for transfer learning and data augmentation. Three Convolutional Neural Network (CNN) models were utilized: the 16-layer Visual Geometry Group network (VGG16), Inception version 4 (Inception_v4), and the 50-layer Residual Network (ResNet50). The dataset used in this research, sourced from Kaggle, contains around 6,400 MRI scans, categorized into four classes: mild dementia, moderate dementia, non-demented, and very mild dementia. A tailored data augmentation pipeline was developed, utilizing techniques such as rotation, flipping, and intensity modifications. This was combined with transfer learning by employing pre-trained models from large-scale image datasets, which were then fine-tuned for AD classification. The performance of the VGG16, Inception_v4, and ResNet50 models was tested under four experimental scenarios: baseline (without augmentation or transfer learning), data augmentation alone, transfer learning alone, and a combination of data augmentation and transfer learning. The findings demonstrated that the integration of transfer learning and data augmentation substantially enhanced classification accuracy, with the top-performing model achieving an accuracy of 98.49%. This method can enhance the accuracy and reliability of AD diagnosis, contributing to more timely intervention and treatment.

Paper Citation

in Harvard Style

Xu D. (2024). Deep Learning Approaches for Early Detection and Classification of Alzheimer's Disease. In Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-754-2, SciTePress, pages 340-348. DOI: 10.5220/0013517400004619

in Bibtex Style

@conference{daml24,
author={Diqing Xu},
title={Deep Learning Approaches for Early Detection and Classification of Alzheimer's Disease},
booktitle={Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2024},
pages={340-348},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013517400004619},
isbn={978-989-758-754-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - Deep Learning Approaches for Early Detection and Classification of Alzheimer's Disease
SN - 978-989-758-754-2
AU - Xu D.
PY - 2024
SP - 340
EP - 348
DO - 10.5220/0013517400004619
PB - SciTePress