White Blood Cells Classification Based on a New Strategy to Cluster Channels RGB of Blood Smear Image Using Resnet-18 Model

Abstract

White Blood Cells (WBCs) play a critical role in the human immune system, and their accurate identification is essential for diagnosing many hematological diseases. Manual analysis of blood smear images (BSI) under a microscope is time-consuming and prone to human error. Recently, deep learning techniques have demonstrated strong capabilities in automated medical image classification. This paper proposes a novel strategy to improve WBC classification by clustering the RGB channels of blood smear images to generate an augmented dataset, which is subsequently used to train a ResNet-18 deep learning model. In the proposed approach, image preprocessing techniques, including morphological operations and RGB channel clustering, are applied to enhance image quality and create additional training samples. The dataset consists of four WBC classes: eosinophils, lymphocytes, monocytes, and neutrophils. Two experiments were conducted to evaluate the effectiveness of the proposed method. The first experiment used the original dataset, while the second experiment used the augmented dataset produced through RGB channel clustering. Experimental results show that the proposed strategy significantly improves classification performance, increasing the overall accuracy from approximately 86.89% and 99.17%. These findings demonstrate that the proposed preprocessing and augmentation approach enhances feature representation and improves the effectiveness of deep learning models for automated WBC classification.