Automated differential white cell counts: a critical appraisal

Abstract

The differential leukocyte count represents a substantial proportion of the workload in routine haematology laboratories. The traditional procedure, in which a technologist scans a Romanowsky-stained blood film and classifies 100 cells, is both labour-intensive and imprecise. The imprecision of the visual DLC is primarily a function of the small number of cells counted. To achieve a significant improvement, however, it would be necessary to count an enormous number of cells, which would be extremely costly in terms of human time and effort. Automation is thus desirable for both economic and clinical reasons. Of the various technological approaches proposed, only two have been developed to the level of marketable devices: digital image processing, and flow cytochemistry. In digital image processing systems, Romanowsky-stained blood films were scanned by a computer-controlled microscope, and leukocytes identified on the basis of parameters analogous to those used by human observers. These systems were relatively slow, offering only a limited degree of automation, and therefore failed to provide significant improvements in terms of workflow and counting statistics. No such instruments are currently marketed. Flow-cytochemical analysers, which classify leukocytes primarily on the basis of size and myeloperoxidase activity, have been available commercially for 15 years, and their clinical utility is well established. A number of haematology analysers offer screening DLCs, providing neutrophil and lymphocyte counts in cases that do not have major abnormalities of the leukocyte population. The performance of these systems depends in large part on their operating environment. Three new automated DLC analysers employing flow cytometric technology have recently been announced, but these have not yet been formally evaluated.