Implementation of Convolutional Neural Networks for the Purpose of Five Types of White Blood Cells Automatic Counting

Grzegorz Dralus, Damian Mazur, Konrad Lukiewicz, Michal Podpora, Jacek Bartman, Henryk Racheniuk, Tomasz Kajdanowicz, Aleksandra Kawala-Sterniuk

Abstract: This study proposes a deep Convolutional Neural Network (CNN) for automated recognition and classification of five WBC types from microscopic images. Various network structures, filter sizes, numbers of hidden layers, and different learning algorithms were evaluated to achieve high accuracy. In this way, 18 network variants, including different learning algorithms, were tested. Several of them achieved very high accuracy in recognizing and scoring 5 types of WBCs. The efficiency of the proposed models can be used to help medical professionals, offering potential support in enhancing diagnostic efficiency and blood analysis.

External IDs:dblp:conf/iccS/DralusMLPBRKK25