Quantitative cytomorphology of megakaryocytes in chronic myeloproliferative disorders--analysis of planimetric and numeric characteristics by means of a knowledge based system

Abstract

Numeric and planimetric parameters of megakaryocytes have been analyzed in 162 bone marrow biopsies of patients with chronic myeloproliferative disorders--CMPD--and controls by means of an inductive knowledge-based system in combination with a multivariate data analysis. To achieve a reliable differential diagnosis between the different entities of CMPD and controls, decision trees and the rank order of the best discriminating parameters have been calculated. The cases measured were defined by 3 histopathologists who were involved in the elaboration of the Hannover Classification of CMPD. The results demonstrate striking numeric and morphologic characteristics of the megakaryopoiesis in each separate primary category of CMPD, that is (1) chronic myeloid leukemia of the common type and (2) with megakaryocytic increase, (3) polycythemia vera, (4) primary or idiopathic thrombocythemia, and (5) chronic megakaryocytic-granulocytic myelosis. Thus, the morphometric measurements did confirm the validity of the Hannover Classification of CMPDs. In order to evaluate the information contained in large quantitative and semiquantitative data bases and diagnostic decisions, knowledge-based expert systems seem to represent a valuable addition to conventional statistics.