Paper Details

Name: Gayathri L , Muralidhara B. L
Country: India

Year: 2025
Volume: Volume-12, Issue-2 (July-December)
Page Number: 48-57
DOI: https://doi.org/10.5281/zenodo.17097235

ABSTRACT:
Classifying Alzheimer’s Disease (AD) using MRI scans is essential for timely detection and
effective treatment planning. This research aims to enhance AD diagnosis by employing
machine learning (ML) models, feature selection methods, and texture-based image analysis.
The study compares the effectiveness of various feature selection strategies and Principal
Component Analysis (PCA) combined with multiple ML algorithms to determine the most
suitable approach for differentiating between Cognitive Normal (CN), Mild Cognitive
Impairment (MCI), and AD cases.
The preprocessing workflow includes N4 bias correction, skull stripping, and linear coregistration, followed by the extraction of texture features that capture statistical
characteristics of image patterns. Different ML classifiers are then trained and tested on these
features to evaluate their ability to accurately categorize patients. Performance is measured
using multiple evaluation metrics to assess the discriminative power of the models across AD
stages.
The findings highlight that combining ML techniques with feature selection and texture
analysis provides a robust framework for early AD detection and personalized treatment
strategies, offering meaningful implications for clinical use.
Keywords— Alzheimer's disease, feature selection, machine learning, texture analysis.