VGG19 and inception V3 performance evaluation for early leukemia detection from blood smears

Abstract

Leukemia is a type of cancer that affects the blood and bone marrow, where the body produces an abnormal amount of white blood cells, leading to impaired immune function. We evaluated two deep learning models, VGG19 and Inception V3, for their performance in detecting leukemia from blood smear images. These models exhibited strong capabilities in distinguishing between benign, early, pre-leukemia, and progressive stages of the disease. The results indicate that both models performed well, with Inception V3 achieving an accuracy of 97.5% and VGG19 achieving 95.5%. Inception V3 showed a slight advantage in early detection sensitivity due to its complex architecture. These findings suggest that these AI-driven approaches have significant potential to enhance the speed, accuracy, and efficiency of leukemia diagnosis. However, further validation with larger datasets is needed to ensure generalizability, contributing to the integration of AI in medical diagnostics, particularly in hematological malignancies.