The study of the effectiveness of classification of images of bone marrow cells in computer systems for diagnostics of acute leukemia and minimal residual disease

The article is devoted to evaluating the model of classification of images of bone marrow cells in the diagnosis of acute leukemia and minimal residual disease using a neural network. The experiment used a sample of 13 cell types: basophils, lymphocytes, monocytes, rod-shaped neutrophils, segmentonuclear neutrophils, eosinophils, lymphoblasts, myeloblasts, prolymphocytes, promyelocytes, normocytes, metamyelocytes, myelocytes. Images of bone marrow cells were obtained from preparations of the Laboratory of hematopoietic immunology of the N. N. Blokhin National medical research center of oncology. The description of cells was performed by twenty-six signs. Models of the used features are presented – the average values of the color components H, S of the color model HSB (H - color tone, S-saturation, B-brightness), morphological characteristics - area, shape coefficient, diameter, the ratio of the maximum distance from the center of mass to the edge of the object to the minimum; textural characteristics of the image area bounded by the cell contour for the spatial adjacency matrix - energy, moment of inertia, entropy, local uniformity, maximum probability for the color components R, G, B, and brightness value. Experimental tests of the classifier under consideration were carried out. The experimental sample contained 636 cells of thirteen different types. It was found that the use of the neural network model for the selected feature system provides 90% accuracy of classification of the studied cell types. The results obtained are of a preliminary nature. An increase in the training sample is required to increase the reliability of estimates in further studies, taking into account the cell types and variability of cell images.

1. Zheng X, Wang Y, Wang G, Liu J. Fast and robust segmentation of white blood cell images by self-supervised learning. Micron. 2018;107:55-71. Epub. https://doi.org/10.1016/j.micron.2018.01.010

2. Acharya V., Kumar P. Detection of acute lymphoblastic leukemia using image segmentation and data mining algorithms. Medical & Biological Engineering & Computing. 2019;57(8):1783–1811. https://doi.org/10.1007/s11517-019-01984-1

3. Mohapatra, S., Patra, D., Kumar, S. et al. Lymphocyte image segmentation using functional link neural architecture for acute leukemia detection. Biomed. Eng. Lett. 2012;2:100–110. https://doi.org/10.1007/s13534-012-0056-9

4. F. Cao, J. Lu, Jianjun Chu, Zhenghua Zhou, J. Zhao and Guoqiang Chen, "Leukocyte image segmentation using feed forward neural networks with random weights," 2015 11th International Conference on Natural Computation (ICNC), Zhangjiajie, 2015:736-742 https://doi.org/10.1109/ICNC.2015.7378082

5. Bhende P.G. Analysis of Blood Cells using Image Processing. International Journal of Science & Technology. 2012;2(3):59-64.

6. Alférez S. et al. Automatic recognition of atypical lymphoid cells from peripheral blood by digital image analysis. American journal of clinical pathology. 2015;143(2):168-176.

7. Rawat J., Bhadauria H., Singh A., Virmani J. Review of leukocyte classification techniques for mi-croscopic blood images. Computing for Sustainable Global Development (INDIACom), 2015 2nd Interna-tional Conference on IEEE. 2015:1948-1954.

8. Wang Q., Wang J., Zhou M. et al. Spectral-spatial feature-based neural network method for acute lymphoblastic leukemia cell identification via micro-scopic hyperspectral imaging technology. Biomedical Optics Express. 2017;8(6):3017-3028.

9. Rawat J., Singh A., Bhadauria H. et al. Clas-sification of acute lymphoblastic leukaemia using hy-brid hierarchical classifiers. Multimedia Tools and Applications. 2017:1-29.

10. Amin M. M. et al. Recognition of acute lymphoblastic leukemia cells in microscopic images using K-means clustering and support vector machine classifier. Journal of medical signals and sensors. 2015;5(1).

11. Linder J., Zahniser D. J. Digital imaging in hematology. MLO: Medical Laboratory Observer. 2012;44(5)

12. Horn C. L., Mansoor A., Wood B. etal. Performance of the CellaVision® DM96 system for detecting red blood cell morphologic abnormal-ities. Journal of pathology informatics. 2015;6.

13. Su M.-C., Cheng C.-Y., Wang P.-C. A neural-network-based approach to white blood cell classification .The Scientific World Journal. 2014;2014.

14. Tay D., Poh C. L., Kitney R. I. A novel neural-inspired learning algorithm with application to clinical risk prediction. Journal of biomedical informatics. 2015;54:305-314

15. Поляков Е.В. Анализ эффективности методов и моделей обработки изображений препаратов крови и костного мозга для автоматизированной диагностики острых лейкозов. Системный анализ и управление в биомедицинских системах. 2019;18(2):133-144.

16. Nikitaev, V.G., Pronichev, A.N., Polyakov, E.V. et al. Application of texture analysis methods to computer microscopy in the visible range of electromagnetic radiation. Bull. Lebedev Phys. Inst. 2016;43:306–308. https://doi.org/10.3103/S1068335616100055

17. Nikitaev V.G., Pronichev A.N., Polyakov E.V. et al. Textural characteristics of bone marrow blast nucleus images with different variants of acute lymphoblastic leukemia. Journal of Physics: Conference Series. 2018;(945):012008. https://doi.org/10.1088/1742- 6596/945/1/012008.

18. Nikitaev V.G., Pronichev A.N., Polyakov E.V., Dmitrieva V. Approach to building knowledge bases in information-measuring systems diagnostics of acute leukemias. Journal of Physics: Conference Series. 2018;(945):012007. https://doi.org/10.1088/1742- 6596/945/1/012007

19. Haralick R. M., Shanmugam K., Dinstein I. Textural features for image classification. IEEE Transactions on Systems, Man, and Cybernetics. 1973; SMC-3: 610-621.