Scholarlink

<p>Alzheimer’s disease (AD) is a cognitive disorder of the brain that leads to gradual decline in neuron density. Electroencephalogram (EEG) signals have been widely used to identify the occurrence of AD in human brain which exhibits reduced brain signals particularly at higher frequency bands. In connection to this, signal variations are widely measured using Hjorth parameters to detect early onset of AD from EEG signals. This work manifests a Discrete Wavelet Transform (DWT) based signal decomposition followed by Hjorth parameter evaluation across various brainwave bands. Afterwards, statistically viable features are extracted through the use of p-test on EEG signals. Proposed methodology is able to come up with minimum number of features which are further utilized for classifying subjects into AD and healthy class using Linear Discriminant Analysis (LDA) and Gaussian Mixture Model (GMM). The performance of different classifiers has subsequently been evaluated. It is shown that Hjorth parameter analysis combined with DWT-based signal decomposition improves upon AD detection accuracy. By using p-test for extracting higher relevance parameter across various frequency bands and the LDA algorithm for classification, highest accuracy of 97.8% is obtained.</p>

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

On the Classification of Wavelet Transformed EEG Signal to Identify Alzheimer’s Disease: A Hjorth Parameter Based Analysis

  • 期刊论文

  • 发表日期：

    2026年4月9日

查看全文
  • Pawan Kumar Ojha,
  • Abhijit Chandra

摘要

Alzheimer’s disease (AD) is a cognitive disorder of the brain that leads to gradual decline in neuron density. Electroencephalogram (EEG) signals have been widely used to identify the occurrence of AD in human brain which exhibits reduced brain signals particularly at higher frequency bands. In connection to this, signal variations are widely measured using Hjorth parameters to detect early onset of AD from EEG signals. This work manifests a Discrete Wavelet Transform (DWT) based signal decomposition followed by Hjorth parameter evaluation across various brainwave bands. Afterwards, statistically viable features are extracted through the use of p-test on EEG signals. Proposed methodology is able to come up with minimum number of features which are further utilized for classifying subjects into AD and healthy class using Linear Discriminant Analysis (LDA) and Gaussian Mixture Model (GMM). The performance of different classifiers has subsequently been evaluated. It is shown that Hjorth parameter analysis combined with DWT-based signal decomposition improves upon AD detection accuracy. By using p-test for extracting higher relevance parameter across various frequency bands and the LDA algorithm for classification, highest accuracy of 97.8% is obtained.