Evaluation of an automated blood smear analyzer.

The performance of a diff3 System prototype instrument that does automated leukocyte differential counts, erythrocytic morphology determinations, and leukocyte count and platelet count estimates, was compared with performance by laboratory personnel for a five-week period in an active general hospital. Using experienced supervisors as the referee method, this instrument performed as well as the routine laboratory staff in all aspects of leukocyte differential counting. With respect to erythrocytic morphology, estimates of hypochromia were inferior to those of the routine laboratory staff; comparisons of estimates of macrocytosis and polychromasia were not significantly different. Leukocyte and platelet numerical estimate performance by the diff3 System was not as good as that of automated cell-counting instruments, but categorization of platelet concentration as normal or abnormal did compare favorably with estimation by laboratory personnel. The flagging of abnormal slides by the instrument was comparable in sensitivity to that by the routine laboratory staff; however, specificity was less, with a significantly greater number of false positives obtained by use of the instrument.