International Journal on Advanced Science, Engineering and Information Technology, Vol. 11 (2021) No. 5, pages: 1847-1853, DOI:10.18517/ijaseit.11.5.9585

Classification of Acute Myeloid Leukemia Subtypes M1, M2 and M3 Using K-Nearest Neighbor

Nurcahya Pradana Taufik Prakisya, Febri Liantoni, Yusfia Hafid Aristyagama, Puspanda Hatta


Leukemia is a malignant disease caused by the massive and rapid development of white blood cells in the bone marrow. These excessive white blood cells begin to interfere with the body’s mechanism rather than fighting infection. Acute Myeloid Leukemia (AML) is one of the four main types of leukemia with eight subtypes, M0 to M7. AML M1, M2, and M3 have similarities, making them more difficult to distinguish from the other types. Furthermore, they are usually identified by calculating the ratio of myeloblast, promyelocyte, and monoblastic. This research aims to apply the k-Nearest Neighbor (k-NN) in classifying these cell types. k-NN is an algorithm used for classification based on a similarity measure. In cases of finding the best number of neighborhoods, trial and error were conducted. The features needed for classification are cell area, perimeter, roundness, nucleus ratio, mean and standard deviation. Four distance metrics such as Euclidean, Manhattan, Minkowski, and Chebyshev were used in this research. The results show that the Euclidean, Manhattan, Chebyshev, and Minkowski distance successfully identified 207 out of 300 objects at K=18, 197 out of 300 objects at K=13,  209 out of 300 correct objects at K=9, and 210 out of 300 objects at K=7.  In conclusion, Minkowski was chosen as the best distance metric for KNN in classifying leukemia-forming blood cells. Furthermore, the accuracy, recall, and precision values of KNN with Minkowski distance obtained from 5-fold cross-validation were 80.552%, 44.145%, and 42.592%, respectively.


Acute myeloid leukemia; classification; K-Nearest Neighbor; white blood cell.

Viewed: 131 times (since abstract online)

cite this paper     download