Measure selection and pattern recognition applied to clinical electroencephalograms

Present techniques of analysis of electroencephalograms (EEGs) require the neurologist to visually scan long segments of the EEG in order to obtain some. diagnostic information. However, much of the information contained in the EEG is lost in the process, and other diagnostic methods are generally required to accurately detect cerebral disorders. The purpose of the study described in this paper is to apply signal processing, measure selection, and pattern recognition techniques to the EEG in order to devise a more accurate and safer method of diagnosis. A particular cerebral disorder, cerebrovascular insufficiency, was used as the basis for the investigation. The analysis of the EEGs can be divided into several stages: recording and digitizing the EEG, making measurements on the EEG, reducing the number of measures to a few representative factors, and analyzing with pattern recognition.