Data Analysis for Comparative Histometry in Pathology - I. Classification Procedures and other Statistical Methods

Abstract

The aim of comparative morphometric procedures in histopathology is to test, whether different groups of patients or different groups of specimen observed can be distinguished or classified based on quantitative morphology. The two predominant developments in quantitative morphology during the last decade include growing spectrum of histometrical methods for the determination of special morphological aspects, as well as a growing spectrum of analytical methods. An overview concerning important methods for data analysis in histometry regarding these developments is thus necessary and will be given in this article. Classification procedures are of primary importance for the analysis of morphometric databases. Different methods, including neuronal networks and knowledge-based systems, will be discussed. According to our analysis especially newer versions of knowledge-based systems provide important possibilities for data analysis and are useful additions to the performance of statistical classification procedures. Implementations and advantages of additional methods for data analysis such as factor analysis for data reduction will be discussed.