Blood Cells Classification for Identification of Acute Lymphoblastic Leukemia on Microscopic Images Using Image Processing

Acute lymphoblastic leukemia (ALL) is a type of leukemia (cancer of the white blood cells) that generally occurs in children. ALL have 3 sub-types, namely Ll, L2, and L3. Microscopic examination to classify ALL subtypes are still done manually by hematologists through visual identification under a microscope, it is difficult to classify ALL subtypes because the characteristics of each subtype are almost the same. This paper proposes a system that is able to detect and classify subtypes of Acute Lymphoblastic Leukemia blood cells using Image Processing. The classification method using K-Nearest Neighbor (K-NN) algorithm based on geometrical and statistical features. In cell object detection, the pre-processing step is used to improve the image quality before going further to the segmentation step using threshold and watershed algorithms. 73 K-NN Dataset from all subtypes of ALL image features were generated to calculate the similarity between new unseen data. In testing results, our proposed classification system achieves 80 % overall accuracy. Each subtype’s accuracy was 75 %, 73.33 %, and 93.33 % for the L1 subtype, L2 subtype, and L3 subtype.