Classification of normal versus malignant cells in B-ALL white blood cancer microscopic images

Abstract

Identification of malignant cells is an important aspect of cancer diagnosis and in determining the degree of progression. Blood disorders, such as leukemia, are most often detected only in advanced stages, when the number of cancer cells is much higher than the number of normal cells. Differentiation of cancer cells from normal blood cells is challenging due to their acute morphological similarity. Early diagnosis of leukemia relies on the accurate classification of malignant cells versus normal cells in microscopic images of blood cells. Computer-aided cell classification has gained popularity as an efficient technique for the diagnosis of leukemia. Our approach to classification involves finetuning of a VGG16 neural network with batch normalization to classify the images of malignant vis-à-vis normal cells.