Journal of Information and Computational Science   2015 年 12 卷 6 期

2015.04.10

In recent year, the five classification for WBC in hematology analyzers are mostly implemented by hardware mode, those instruments are excessively rely on the accuracy of some components, and the hardware structure is complex, which limit the further development of the five classification hematology analyzer. Therefore, a five classification system for WBC based on RBF neural network is put forward, which takes the full-optical technology as the WBC detection method, and uses VC6.0 as the software development platform to establish the RBF neural network model for the recognition and five classification of WBC. The experimental results show the recognition accuracy of the instrument equipped with the system is close to Mythic 22 which is a high-grade five classification hematology analyzer. Conclusions: The five classification system for WBC proposed has the features of reliable performance and high degree of accuracy.