Date of Award
Summer 1991
Document Type
Thesis - Restricted
Degree Name
Master of Science (MS)
Department
Biomedical Engineering
First Advisor
Myklebust, Joel
Abstract
Early detection and subtyping of dyslexic readers may relieve the suffering and frustration experienced by the dyslexic readers and families. Earlier diagnosis means an earlier start in therapy for patients. Dyslexia is usually diagnosed by subjecting patients to empirical analysis such as submission of achievement and cluster analysis, or through clinical evaluation of psychological processes which stresses reading development. Even though these tests are essential for dyslexic diagnosis and subtyping, the diagnosis is a subjective one left to the psychologist. Researchers, in their quest for a better, efficient, and less subjective method for identifying and understanding dyslexia, have turned their attention towards electroencephalography for answers. In understanding dyslexia researchers are interested on whether dyslexia is a neural deficit or is dyslexia a result of poor learning or teaching. However, the electroencephalogram of a dyslexic subject does not reliably differ from that of a normal subject on visual inspection. Electroencephalograms were then quantified for dyslexic identification and subtyping. The present work is a further step along the road to analyze quantified electroencephalogram to identify dyslexic subtypes. Neural networks, a mode of artificial intelligence, is used to classify dyslexia on the basis of quantitative electroencephalogram; several neural nets were constructed to reach for an optimal neural net. The Introduction offers a brief description of methods involved in quantifying the electroencephalogram. Chapter I describes the physiology of the biological neuron, the mechanisms of the Central Nervous System, the origins of potentials measured in the electroencephalogram, and their significance. The objective of this research and a brief discussion of the history of neural networks and artificial intelligence is also introduced in chapter I. Chapter II introduces the fundamentals necessary for the use and understanding of neural nets. Chapter III is composed of two sections: methods and results. Chapter IV discusses the results obtained and their significance in electroencephalogram analysis. Also, the conclusion derived from this research and implications for future work are included in chapter IV.
Recommended Citation
Ramadan, Zahi Hussein, "A Neural Network Approach to Discriminate Between Normal and Dyslexic Subtypes in Children" (1991). Master's Theses (1922-2009) Access restricted to Marquette Campus. 3969.
https://epublications.marquette.edu/theses/3969