Unsupervised Method for Calculating Diameter and Number of Leukocyte Cells

Abstract

Leukocyte has an essential role in the body. Leukocyte has functioned as the body's immune system function to protect the body from viruses, bacteria, and pathogens. Under normal circumstances, leukocyte has a relatively constant amount. If the number of leukocytes exceeds or is less than usual, health problems can occur. So that the leukocyte cell counts as an indicator of abnormalities and diseases. Digital image processing is an alternative method to identify leukocyte in addition to a hematology analyzer or manually by medical personnel by processing blood images so that the desired information is obtained. Blood images have a variety of lighting conditions, staining time, and cell thickness that cause the various attributes of the blood image. This experiment segmented leukocyte using unsupervised method especially the K-Means clustering method on the hue component of the image after being transformed into HSV color coordinates by grouping leukocyte cell images (great hue value) into one cluster. The results show that the performance of our system as follows, sensitivity 84%, specificity 94%, and accuracy 97%.