Optimizing White Blood Cell Classification with ResNet50 Neural Network

In medical diagnostics, the categorization of white blood cells (WBCs) is a crucial activity necessary for diagnosing and surveilling many health issues such as immunological diseases, infections, and hematologic malignancies. This paper investigates automatic WBC categorization using deep learning methods, namely the ResNet50 model. ResNet50 is a 50-layer residual network well-known for its deep architecture and residual learning capabilities, enabling it to learn complex patterns and characteristics. The ResNet50 model has been fine-tuned to differentiate between neutrophils, eosinophils, basophils, monocytes, and lymphocytes, among other WBC types. To improve generality and resilience, our method included careful model training, augmentation, and data preprocessing. Standard measures including accuracy, precision, recall, and F1-score were used to assess the model's performance, which showed that it was effective in obtaining high classification accuracy and reliability. The findings show that WBC classification accuracy and efficiency are greatly increased by the ResNet50 model, which may have advantages for clinical practice. The approach facilitates faster and more precise diagnosis by pathologists by automating the categorization process. This article uses the ResNet50 model, which has a 95% accuracy rate in balanced precision, recall, and f1-scores for categorizing white blood cells.