Automatic Recognition of Leukemia Cells using Texture Analysis Algorithm

A trained laboratory specialist usually scored the detection of leukemia cells by determining the leukemia location and size subjectively (by eyes). This subjective method will add to the 5% tolerance error, which might compromise the whole process of treatment especially in patients with severe conditions. The aim of this study is to increase the edge recognition in leukemia cells images in patients with leukemic disease automatically using L*a*b* color space and K-means clustering. First, we read the microscopic images. We then to convert the images form RGB color space to L*a*b* color space. Then we classify the colors in 'a*b*' space using K-means clustering. Then we label every pixel in the Image using the results from K-means. We then create images that segment the leukemia image by color. Finally, we segment the leukemia cell image into a separate image. The sample of this study was (46 cases) and they showed increase enhancement. This segmentation technique (automatic scoring) and segmented images was adjudicated by trained laboratory specialists as being comparable to other segmentation techniques created with manual editing (subjective scoring. The quantitative results calculated using a measure of percentage match between ground truth and segmentation results. The percentage match (PM) measure was 99.33 (p ?0.05) and Corresponding Ratio (CR) was -0.007 p ?0.05).