Novel Deep Neural Network Model for Computer Vision-Based Leukemia Detection

Acute lymphoblastic leukemia (ALL) is a common subtype of leukemia that requires prompt and precise diagnosis. Leukemia is characterized by enhanced white blood cells in the bone marrow or peripheral blood. Using peripheral blood smear pictures, this paper suggests an AI-driven method for classifying ALL subtypes: Benign, Early Pre-B, Pre-B, and Pro-B. Sophisticated preprocessing methods improved training image quality. For early leukemia identification, this approach provides a quick and particularly in environments with limited resources. NASNetMobile, EfficientNetB0, and MobileNetV2 were among the lightweight pre-trained CNN models used. With an accuracy of 99.62 percent, MobileNetV2 outperformed EfficientNetB0 and NASNetMobile.