Advanced Machine Learning for Leukaemia Detection Based on White Blood Cell Segmentation
Cancer is a worldwide issue right now. It is generated from blood cells. In fact, the diseases is spread by a change in WBC cells that can cause an unspecified growth in the number of abnormal cells. These immature cells start disrupting the function of n
- PDF / 1,098,059 Bytes
- 13 Pages / 439.37 x 666.142 pts Page_size
- 1 Downloads / 145 Views
and Manzoor Ahmad Chachoo
University of Kashmir, Srinagar 190006, India [email protected], [email protected]
Abstract. Cancer is a worldwide issue right now. It is generated from blood cells. In fact, the diseases is spread by a change in WBC cells that can cause an unspecified growth in the number of abnormal cells. These immature cells start disrupting the function of normal cells. Processing medical blood images in ML follows a stepwise procedure from pre-processing of blood components to subtype classification. Blood components discrimination, WBC cell identification, noise reduction and cell counting is done by choosing the methodology for segmentation. Its accurate implementation is difficult because of the varying parameters (Size, Eccentricity, Major Arc, Minor Arc and Parameter) of the different WBC types. Usually segmentation is done by extracting nucleus of the WBC cells. Our research is the first that have found some new problems (Division of WBC Nucleus into multiple parts, Distortion in the shape of WBC cell Nucleus, Size Variation and Noise in the form of pixel scratches) associated with the ML methods that remove cell cytoplasm with other blood components (RBC, Platelets and Background) during the nucleus extraction process of WBC cell. Parallelly we have found causes for the mentioned problems, to suggest generic methodology that will produce robust results. Experimental results have shown that methods extracting Nucleus of the WBC cell works only with a certain type of blood images. When we try to generalize the methodology for different types of blood cell images, we are getting the above-mentioned problems. The results had been calculated by the experiments done in MATLAB. Keywords: ALL-IDB (Acute lymphoblastic Leukaemia – International Database) ML (Machine-Learning) DL (Deep-Learning) WBC (WhiteBlood-Cell) RBC (Red-Blood-Cell)
1 Introduction White blood cells (WBC) are the main component of our immune system. They originate from bone marrow and protects our body from different types of diseases [1]. In human body we have different types of WBC (Lymphocyte, Monocyte, Neutrophil, Basophil and Eosinophil) cells and each subtype have different size and curvature as shown in Fig. 1 to perform normally human body needs specified number or range of © Springer Nature Singapore Pte Ltd. 2020 C. Badica et al. (Eds.): ICICCT 2020, CCIS 1170, pp. 195–207, 2020. https://doi.org/10.1007/978-981-15-9671-1_17
196
I. M. Sheikh and M. A. Chachoo
values from each subtype of WBC. It can damage huge to a human health if there is a slight deviation from over or under the normal limits in the value of cells. Research in medical field have shown different types of diseases in the form of bacterial infections, Allergy, AIDS, CLL, and ALL grow when the varying range of WBC types present in the hemoglobin met a particular threshold. Among the mentioned diseases associated with different sub-classes of WBC’S cancer has uncontrolled growth in the number of blast cells generated from bone
Data Loading...
