MEASURE SELECTION AND PATTERN RECOGNITION IN THE EEG.
TEXAS UNIV AUSTIN LABS FOR ELECTRONICS AND RELATED SCIENCE RESEARCH
Pagination or Media Count:
The purpose of this research is to determine if computer analysis of the electroencephalogram EEG could be used to detect Vitamin B6 deficiency in chicks. The existence and extent of diet deficiency has traditionally been measured by weight and behavioral characteristics. Although these measures very accurately identify the diet deficiency they do not provide for its early detection. It is the goal of this research to develop a technique that will detect the deficiency syndrome using only measures extracted from the EEG before the syndrome is identified by the traditional measures. The first step in the procedure is to select the set of all EEG measures which contain information related to Vitamin B6 deficiency. The second step is to select a small subset of these measures. The subset should have the greatest probability that a pattern recognition algorithm using it will accurately classify the chicks as belonging to either the control or deficient group. The relative quality of the measure selection procedure will be determined by one of two criteria 1 whether it produces the minimum error, or 2 whether it allows the earliest detection. The method which is used in this research for selecting the measures may be extended to other problems which involve time varying signals. In order to use this procedure it would be necessary for the new problem to have a structure similar to that of the chick deficiency problem. Author
- Medicine and Medical Research
- Biomedical Instrumentation and Bioengineering