Back to search

Clemson University

The Development of an Automatic EEG Analysis System Utilizing an Interactive Feature Extraction Program Based on Modified HAAR Coefficents

Abstract

dc:description.abstract

There is a need for an automatic analysis system for electroencephalograms. A reliable and inexpensive system could supplement the present medical personnel to significantly increase the utilization of the valuable medical tool. This research contributes to the attainment of this generalized analysis system by, (l) developing the fundamental understanding, the methodology, and the computer programs nee ded for a long range study, and (2) providing a cornputer program that reliably deterntines the EEG sleep stage . The primary contribution to the long range study problem is the user-oriented, feature extraction program, HAARlS. This is an interactive program ·written for the DEC PDP/ 15 computer that inputs a single channel of data and outputs modified Haar coefficients or functions of the coefficients. This program allows a researcher to easily analyze, with modified Haar coefficients, an EEG, or any other signal. The possible uses of the system were demonstrated by analyzing three signals, a pulse-in-noise test signal, textile-yarn mass data, and EEGs. The automatic sleep staging system was developed to clearly establish the value of the modified Haar coefficients in pattern recognition as well as to provide a system to reduce the work load on the electroencephalorapher. The program analyzes one EEG channel and automatically determines the sleep stage from among the possibilities of Stage W and Stages 1 through 4. As an electro-oculogram is not used, the REM Stage is not recognized. The program was tested by analyzing three all-night records with the sections of REM Stage deleted. The automatic sleep decisions agreed with electroencephalographers decisions 88.7% of the time.

Degree

thesis:*
Name thesis:degree_name
Doctor of Philosophy (PhD)
Level thesis:degree_level
Dissertation
Discipline thesis:degree_discipline
Electrical and Computer Engineering

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Barnett, Jackson William
Contributors dc:contributor
  • A. E. Schwertz

Identifiers

dc:identifier.*
Repository record dc:identifier
https://open.clemson.edu/arv_dissertations/8

Chain of custody

source
Harvested from
Clemson University
Base URL
open.clemson.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Barnett, Jackson William. The Development of an Automatic EEG Analysis System Utilizing an Interactive Feature Extraction Program Based on Modified HAAR Coefficents. Dissertation thesis, https://open.clemson.edu/arv_dissertations/8