Discriminative Power of Lymphoid Cell Features: Factor Analysis Approach

摘要

The new results of the research in the field of automation of hemato-poietic tumor diagnostics by analysis of the images of cytological specimens are presented. Factor analysis of numerical diagnostically important features used for the description of lymphoma cell nucleus was carried out in order to evaluate the significance of the features and to reduce the considered feature space. The following results were obtained: a) the proposed features were classified; b) the feature set composed of 47 elements was reduced to 8 informative factors; c) the extracted factors allowed to distinguish some groups of patients. This implies that received factors have substantial medical meaning. The results presented in the paper confirm the advisability of involving factor analysis in the automated system for morphological analysis of the cytological specimens in order to create a complex model of phenomenon investigated.