Automatic classification of leukocytes

Abstract

This paper presents a novel algorithm for the automatic compilation of differential blood count (DBC), which is based on the direct analysis of a blood smear image and artificial neural networks. The results of the algorithm testing show high sensitivity of the algorithm in leukocyte detection and classification accuracy of 86%. Also, the algorithm enables the detection of potentially falsely classified leukocytes and in that way, with the help of a hematological expert, enables additional increase in the DBC compilation quality.nema