“Event related potentials, or ERPs, are averaged waveforms of EEG cerebral activity measured using non-invasive electrodes attached to the scalp. They occur at a particular time before, during, or after some physical or psychological event. Certain diseases resulting in neuropsychological impairment seem to result in changes in the waveform and latency of the P300 auditory evoked potential. It would be desirable to find a method to examine human ERP measurements with the goal of detecting the presence of and measuring the progress of such diseases.
The detection of multiple sclerosis using such a method would allow for the detection of this disease without resorting to a battery of complex tests that would need to be administered by highly trained specialists. If the abnormalities exist before other symptoms become evident, such a detection method might benefit the prognosis for patients, as well as providing a benchmark for measuring the progression of the disease.
Experiments using artificial neural networks to analyze ERP data sets have been performed in previous research by Slater, Wu, Ramsey, Honig and Morgan. Using feed forward backpropagation neural networks, those researchers obtained classification accuracies of 80% for single channel (electrode) training. By constructing a simple voting jury that assigns a classification based on a vote of 2 out of 3 single channel outputs, accuracies of 90% were achieved. This research describes a series of experiments on the same data sets undertaken with the goal of seeking higher accuracies. Single channel training using inputs that were pruned to isolate the P300 resulted in classification accuracies approaching 86%. By constructing a multi-channel ensemble network that used raw outputs of the single channel networks as inputs, classification accuracies of 96% were achieved”--Abstract, page iv.
St. Clair, Daniel C.
Wilkerson, Ralph W.
Dagli, Cihan H., 1949-
M.S. in Computer Science
University of Missouri--Rolla
ix, 48 pages
© 2001 Greg Eastman, All rights reserved.
Thesis - Restricted Access
Print OCLC #
Link to Catalog Record
Electronic access to the full-text of this document is restricted to Missouri S&T users. Otherwise, request this publication directly from Missouri S&T Library or contact your local library.http://merlin.lib.umsystem.edu/record=b4728512~S5
Eastman, Gregory Charles, "Detection of multiple sclerosis using neural network ensembles" (2001). Masters Theses. 2087.
Share My Thesis If you are the author of this work and would like to grant permission to make it openly accessible to all, please click the button above.