European Journal of Molecular & Clinical Medicine

Abstract

The death caused by Leukemia has been ranked in the top ten most dangerous mortality cause for the human being. There are numerous reasons and causes, in spite of the causes and reasons the profound problem is the slow decision-making process which delays the time required to proceed with medical treatment for the patients. That’s why the enhanced medical support process has become necessary for the classification of leukemia. The four different types of Leukemia are as follows Myeloid Leukemia where we have acute and chronic subcategories and in the same way, it goes for the myeloid type as well, these affect various cells and systems such as the blood cells, bone marrow, lymphatic system and which causes the death of patients. The proposed method improves the CML, CLL, AML and ALL characteristic accuracy by scanning color and textural features from the blood image using image processing and to aid in the grouping of CML, CLL, AML and ALL. The following technique proposes a quantitative microscopic approach toward the grouping of blood sample images. A model using Modified Convolution Neural Network (CNN) architecture is used to optimize the classification process. Based on optimized feature space, a CNN model with various kernel functions (filters) used to abstract the features from the pixel values. The proposed method is tested using nearly 10000 microscopic blood images. The outcome confirmed that the accuracy of the classification using blood sampled images which was up to 98%.