Cells identification of acute myeloid leukemia AML M0 and AML M1 using K-nearest neighbour based on morphological images

Abstract

Acute Myeloid Leukemia (AML) is a type of leukemia characterised by the occurrence of myeloid series cell differentiation that stops in the blast cells causing the accumulation of blast cells in the bone marrow. This study aims to determine leukemia typically in AML M0 and AML M1 based on the morphology of white blood cell image using image processing method. The steps performed are median filtering, YCbCr colour conversion, thresholding, and opening, and k-Nearest Neighbors classifier to classify cell types from feature extraction results. The result of characteristic extraction was done by mean difference test for each characteristic between cell type indicated that there was a significant difference in WBC diameter characteristic between cell type, while on a characteristic of nucleus ratio showed that there was no significant difference. Based on characteristic testing of each cell, a combination of a characteristic of WBC diameter and nucleus roundabout obtained the highest accuracy when k = 5 and k = 7 is 67,28%. Thus the characteristic of WBC diameter and the nuclear roundabout is the most influential data classification feature. Based on the test results of each cell, if the algorithm k = 6 k-Nearest Neighbors can classify the cell correctly 59.87% of the 162 data used based on the three characteristics each cell is the WBC diameter, the nucleus roundabout and the nucleus ratio.