A Comparative Analysis of CNN, XG Boost, DNN and Res Net for Leukemia Classification Using Hybrid PSO Model

Major health concern is leukemia, a form of blood cancer marked by aberrant white blood cell development. Effective treatment depends on an early and precise diagnosis, and deep learning (DL) has become a game-changing tool in this field. A thorough examination of several deep learning models, such as ResNet, XG Boost, Convolutional Neural Networks (CNNs), and Deep Neural Networks (DNNs), is given in this review study. For prompt treatment, leukemia must be identified from peripheral blood smear images as soon as possible. Although medical image analysis is dominated by Convolutional Neural Networks (CNNs), recent research has investigated ensemble and gradient boosted techniques as XG Boost, Deeper ResNet, and fully connected DNNs. Four typical models—custom CNN, ResNet 50, feature fusion with XG Boost, and a multilayer DNN—are experimentally compared head-to-head on the public ALL IDB1, ALL IDB2, and C NMC 2019 datasets in this study. ResNet 50 had the highest macro F1 (96.8%), followed by CNN (94.3%), DNN (90.1%), and XG Boost (88.5%). The significance of ResNet's margin (p