A Novel CNN Model with Entropy-Coded Genetic Algorithm for Blood Cell Classification

Abstract

White blood cells are the immune system components that combat infections. White blood cells or WBCs, are blood cells that are present in the bone marrow are responsible for protecting against pathogens that kill healthy cells. Finding the immature cell formations early on will help to lessen the severity of this problem and eventually reduce the patients' rate of modality. In this study, an entirely new deep Convolutional Neural Network (CNN) model is presented inside this paradigm. The deep CNN model is pre-trained on medical imaging datasets to improve its performance. In preprocessing, K-Mean Clustering is utilized to highlight the region of interest (ROI). Post clustering, the updated dataset is used for feature extraction using a novel proposed CNN model with 33 layers. Moreover, the Entropy-Coded Genetic Algorithm is another significant contribution which is used for feature selection to choose the most optimal features. These selected features are subsequently classified using a Support Vector Machine (SVM). The results show that with 2025 features, the proposed model achieved 97.59% accuracy, 96.0% sensitivity and 97.91% specificity using the cubic-SVM classifier.