Classification of normal and abnormal samples of peripheral blood by linear mapping of the feature space

摘要

A data set of features derived from digitized images of monocytes and lymphocytes taken from samples of normal persons (10 cases) and from persons with either Hodgkin's disease (10 cases) or non-Hodgkin's lymphoma (9 cases) was explored. The morphological characteristics of a small proportion of the cells in the abnormal samples may well provide indications of the abnormality.A linear mapping of the feature space is described. This technique may be used to visualize the abnormality of the sample and to design a decision rule, applicable to independent test samples.