Özet:

White blood cells are the most important structure of the human immune system and are produced by the bone marrow in the blood and lymph tissues. These cells are the defense mechanisms that protect the human body against diseases and foreign organisms, and Leukopenia may be encountered when the rate in the blood decreases. Intensive effort is required to determine the ratio of these cells in the human body and to detect and treat the disease. In this study, it is aimed to improve the white blood cell classification performance by using deep learning models. Convolutional neural network models are used to perform the classification process more efficiently. Densenet201, ResNet50, and Alexnet were combined to distinguish between the white blood cell variants, eosinophils, lymphocytes, monocytes, and neutrophils. Three different machine learning classifiers, namely K-Nearest Neighborhood, Support Vector Machine and Naïve Bayes, were used to classify the obtained feature maps. By applying CLAHE and Gaussian filters to Kaggle dataset images trained with Deep Learning (DL), these images were reclassified with three CNN networks. It has been revealed that the results obtained after applying these filters are higher than the classification results obtained with the original data.

Anahtar Kelimeler: