Computer-Aided Diagnosis of White Blood Cell Leukemia using VGG16 Convolution Neural Network
G. Meena Devi, V. Neelambary · 2022 4th International Conference on Inventive Research in Computing Applications (ICIRCA) · 2022
Leukemia is a type of blood cancer caused by the uncontrolled multiplication of white blood cells (WBCs) within the bone marrow. This kind of white blood cell (WBC). Current Convolution Neural Networks are the best solution for medical image analysis, such as detection and classification (CNN). The structure AlexNet, VGG16, GoogLeNet, and ResNet50 was evaluated with complete learning and transfer learning when training CNN, which is the cornerstone of the Convolution layer. The technique successfully determined WBC cells 100 percent of the time at the end of the research. With transfer learning, VGG16, one of the CNN designs, has shown the greatest results. Lymphocyte cell types were determined with a 98.53 percent accuracy rate, Monocyte cell types were determined with a 97.41 percent accuracy rate, Basophil cell types were determined with a 98.48 percent accuracy rate, Eosinophil cell types were determined with a 96.16 percent accuracy rate, and Neutrophil cell types were determined with a 94.05 percent accuracy rate. On each of these VGG16 datasets, their results were some of the best, outperforming some convolutional network models. The average accuracy of the methods applied in image analysis to detect leukemia was 97.16%.