Clinical application of cytomorphological analysis based on artificial intelligence in leukocyte classification

Objective To explore the application value of automatic cytomorphological analyzer in the morphological analysis of white blood cells (WBC) in peripheral blood.

Methods A total of 306 venous blood samples from inpatients and outpatients were randomly selected and prepared with automatic cytomorphological analyzer for WBC pre-classification. The differences between automatic cytomorphological analyzer counting, automatic blood cell analyzer counting and manual counting were compared, and the correlation between automatic cytomorphological analyzer and manual counting method was analyzed.

Results Compared with the other two methods, the automatic cytomorphological analyzer was able to detect more types of WBC, especially abnormal cells. There were no significant differences between automatic cytomorphological analyzer and manual counting method for 6 mature WBC types (band neutrophils, segmented neutrophils, lymphocytes, monocytes, eosinophils, and basophils), immature cells at different stages and atypical lymphocyte counts (P>0.05). Results of the 6 mature WBC types counted by the automatic cytomorphological analyzer and manual counting had favorable correlations (r>0.8).

Conclusion The automatic cytomorphological analyzer can classify more types of WBC, provide WBC counting results that are highly consistent with manual microscopy, and the counting results of the two methods have a good correlation.