Abstract

This research deals with the study of Alzheimer Disease (AD). Electroencephalogram (EEG) signal is a clinical tool for the diagnosis and detection of AD. EEG signals are analyzed for the diagnosis of AD applying several linear and non-linear methods of signal processing. This work studies and implements several measures of EEG signal complexity and then compares the complexity features measured or extracted from EEG signals. Time domain analysis of EEG signals is performed using several signal processing techniques such as higher order moments, entropies and fractal dimension calculation using fractal analysis. Frequency domain analysis of EEG signals is performed using signal processing techniques such as Welch Power spectrum and Discrete Fourier Transform (DFT). EEG signal analysis using Wavelet Transform was also performed. Higher order moments, entropies, fractal dimension estimation using fractal analysis and Welch Power Spectrum are also implemented along with moving windows. This work also deals with the artifact removal or de-noising of EEG signals using a band pass filter. EEG signal data recorded from AD subjects and their respective age-matched control subjects are used to test the performance of the methods in diagnosing AD. In addition, this work outlines the drawbacks of the methods used and compares the methods for the best feature extraction techniques.

Date of publication

Spring 4-30-2012

Document Type

Thesis

Language

english

Persistent identifier

http://hdl.handle.net/10950/65

Download

Share

COinS