Search

The report describes experiments set up to evaluate the relative merits of period and frequency analytic methods and to compare the results with visual reading and intepretation of clinical electroencephalograms. The investigation advanced through three phases. The first phase dealt with both frequency and period analyses by the condenser method of wave sampling. The second phase utilized a 200 per second full-wave zero cross sampling. The third phase employed a half-wave zero cross sampling method. Statistical analysis included clinical evaluation vs period counts and frequency histograms, analysis of variance with five-group and pairwise comparisons, discriminant analysis and clustering. Period analyses compares favorably with clinical interpretation and frequency analysis. A minimum of 6 variables selected from the 72 total sample variables in period analysis are deemed sufficient to derive adequate computer discrimination between various groups of patients.