Pattern Recognition Analysis of Ascitic Fluid Chromatograms for Patients with Liver Disorders

Abstract Pattern recognition techniques have been applied in a number of medical applications. In the method described here, new orthogonal polynomials developed by Cohen are used as potential functions in a supervised learning approach to pattern recognition. This method is applied to the analysis of chromatograms obtained from ascitic fluid taken from patients who have spontaneous bacterial peritonitis, as well as from a control group. The chromatograms show the presence of organic acids in the ascitic fluid. The objective of the pattern recognition analysis is to obtain a method of identifying patients with spontaneous bacterial peritonitis by pattern analysis of the chromatogram. Results obtained show that as high as ninety-seven percent of the cases may be classified correctly using this method. The pattern recognition method is also compared to standard statistical discriminant analysis techniques, and is shown to be superior.