ENHANCED DEEP CNN FOR PRECISE MULTICLASS CLASSIFICATION OF BLOOD CELL SUBTYPES IN HISTOLOGICAL IMAGES

Abstract

White blood cells (WBCs), or leukocytes, are a crucial component of the immune system, responsible for defending the body against infections through mechanisms such as pathogen ingestion, antibody production, and direct destruction of infectious agents. Leukocytes are classified into five major subtypes: neutrophils, eosinophils, lymphocytes, monocytes, and basophils. Accurate identification and quantification of these subtypes play a vital role in clinical diagnosis, aiding in the detection of infections, immune disorders, and the monitoring of cancer treatments. Traditional manual differentiation of WBCs under a microscope is labor-intensive, time-consuming, and prone to errors, highlighting the need for automated classification techniques.Conventional machine learning-based classification methods rely on handcrafted features and often involve multiple steps, including preprocessing, segmentation, feature extraction, and classification, which may compromise accuracy and efficiency. In contrast, deep learning, particularly convolutional neural networks (CNNs), has revolutionized image-based classification by enabling end-to-end learning from raw images. This study presents an Enhanced Deep CNN for precise and automated multiclass classification of blood cell subtypes from histological images. The proposed approach leverages advanced CNN architectures to learn hierarchical feature representations, improving classification accuracy and robustness. By eliminating the dependency on manual feature selection, this model enhances diagnostic efficiency, offering a powerful tool for hematologists and medical practitioners in WBC analysis