Leukemia Classification Using CNN and VGG19 Architecture

Leukemia, a severe kind of blood cancer marked by aberrant white blood cell proliferation, presents great difficulties for diagnosis and categorization. Effective treatment planning and enhancement of patient outcomes depend on precise identi- fication of leukemia subtypes. Based on the VGG19 architecture, this paper offers a Convolutional Neural Network (CNN) model to classify leukemia into four subtypes: Benign, Malignant Pre- B, Malignant Pro-B, and Malignant early Pre-B. High accuracy is obtained by the model using transfer learning and fine- tuning on a dataset of leukemia cell images, thereby lowering training time and improving performance. Our results show that, surpassing conventional diagnostic approaches and current Machine Learning (ML) models, the VGG19-based CNN model achieves an accuracy of 99.1%. This study underlines the need of automated systems in the diagnosis process and the possibilities of Deep Learning (DL) methods in medical image classification. Aiming for enhanced accuracy and efficiency in clinical settings, the results open the path for more research of sophisticated computational methods in the diagnosis of leukemia.