ABSTRACT

MULTIRESOLUTION WAVELET ANALYSIS OF EVENT RELATED POTENTIALS FOR THE DETECTION OF ALZHEIMER'S DISEASE

by

ROBI POLIKAR

Abstract


Alzheimer's disease is a neurological disorder (dementia), 
usually seen in elderly people, which can only be diagnosed 
(for sure) via an autopsy, i.e., after the patient has died.  
Neurologists can usually diagnose Alzheimer's  disease from 
symptoms, however, misdiagnosis is quite common.  Although 
Alzheimer's cannot be cured, early detection of the
disease can help in delaying the progress of the disease.   

In this study, we have tried to diagnose the disease by 
analyzing the electroencephalograms (EEG) of  the patients.
The signals acquired with the "oddball paradigm" experiment 
were analyzed using multiresolution wavelet analysis techniques. 

Multiresolution wavelet analysis is a relatively new technique
in signal processing, especially suited for the analysis of 
non-stationary signals. It offers a represention of signals in 
the time-frequency plane, giving information regarding the time
localization of the spectral components.  EEG signals are highly 
non-stationary in nature, and therefore, they are not suited for
Fourier transform based analysis.  

The analyzed signals were then used to train a neural network.
Both supervised and unsupervised learning rules were used to 
distinguish the signals that belong to patients with Alzheimer's 
disease from the ones that belong to control patients without 
Alzheimer's disease. Initial results are very promising, and 
demonstrate the feasibility of this approach.

There has been very little study on early detection of Alzheimer's
disease by using signal processing techniques, and we believe that
future study on this approach will result in more reliable diagnosis of
Alzheimer's disease.




I am currently investigating the possibilities of providing the
the complete thesis from this homepage, but I am having some 
problems due to its massive size. If you are interested, please 
drop me a line, and I will do whatever I can.

Please send any comments and suggestions by using the mailform.
Just click here to  send me an e-mail. 

Back to Robi Polikar Cyberdomain Root Page
Wavelet Tutorial Main Page