Scholarlink

Alzheimer’s disease (AD) causes a progressive decline in cognitive abilities that affects memory, thinking, and behavior. Functional Magnetic Resonance Imaging (fMRI) has emerged as a valuable tool in this process, enabling the study of neural activity and connectivity patterns in the brain. Efficient analysis of resting-state fMRI data is essential to identify connectivity disruptions that suggest early signs of dementia. In the present work, we propose a machine learning-based system designed to detect AD through the analysis of resting-state fMRI data. The proposed pipeline involves preprocessing fMRI data, constructing functional connectivity networks, and evaluating the performance of multiple machine learning algorithms. Six classifiers were built and analyzed: four traditional and two deep learning. Computational experiments were conducted using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The results demonstrate that traditional machine learning outperforms deep learning for the detection of AD. Among the models evaluated, the Support Vector Machine achieved the highest test accuracy of 97.37%, significantly outperforming the ResNet-18 deep learning architecture, which obtained an accuracy of 71.05%. These findings underscore the effectiveness of traditional machine learning methods in the context of limited datasets.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

fMRI Analysis for Alzheimer’s Disease Detection: Traditional vs. Deep Learning Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Adél Bajcsi,
  • Camelia Chira

摘要

Alzheimer’s disease (AD) causes a progressive decline in cognitive abilities that affects memory, thinking, and behavior. Functional Magnetic Resonance Imaging (fMRI) has emerged as a valuable tool in this process, enabling the study of neural activity and connectivity patterns in the brain. Efficient analysis of resting-state fMRI data is essential to identify connectivity disruptions that suggest early signs of dementia. In the present work, we propose a machine learning-based system designed to detect AD through the analysis of resting-state fMRI data. The proposed pipeline involves preprocessing fMRI data, constructing functional connectivity networks, and evaluating the performance of multiple machine learning algorithms. Six classifiers were built and analyzed: four traditional and two deep learning. Computational experiments were conducted using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The results demonstrate that traditional machine learning outperforms deep learning for the detection of AD. Among the models evaluated, the Support Vector Machine achieved the highest test accuracy of 97.37%, significantly outperforming the ResNet-18 deep learning architecture, which obtained an accuracy of 71.05%. These findings underscore the effectiveness of traditional machine learning methods in the context of limited datasets.