Within the realm of illness diagnostics, one of the most significant challenges is the early detection and diagnosis of leukemia. In order to successfully overcome this obstacle, it is necessary to accurately differentiate between healthy and malignant leukocytes during the early stages of the disease while simultaneously minimizing costs. Leukemia is a disease that affects a large number of people, yet there are only a few flow cytometers available, and the diagnostic processes that are carried out in laboratories are time-consuming. This is accomplished by contrasting three models, notably the regular CNN model and the deep CNN model (Alex Net and the VGG-16 Net model). As indicated by the C-NMC 2019 dataset, a total of 11,154 blood microscopic images were gathered for the aim of analyzing the approach that we have proposed. It has been observed, on the basis of the findings of our research, that the performance of a VGG-19 Net model is superior to that of other two models, such as the Traditional CNN model and the Alex Net model. By employing the VGG-19 Net model as a feature extractor and Soft-max as the classifier, the model is able to attain the highest possible level of performance. With this arrangement, the accuracy is 97.44, the precision is 97.5, the recall is 97.5, and the F1-score is 97.5.