LPG-PCA algorithm and selective thresholding based automated method: ALL & AML blast cells detection and counting

In this paper, an automated method is designed to detect Acute Lymphoblastic Leukemia (ALL) and Acute Myeloid Leukemia (AML) blast cells from human microscopic blood cell images. The developed method comprises of four basic modules. The de-noising module performs two staged noise reduction using Principal Component Analysis (PCA) and Local Pixel Grouping (LPG). The contrast enhancement section includes colour space conversion and morphological filtering based on pixel intensities. In threshold selection module, threshold value is determined using two methods, namely, Edge sensitive Variational Thresholding and Otsu´s Thresholding. Blast cells are segmented based on threshold value obtained from these two methods. Morphological operations and Connected Component Analysis are used to count the number of blast cells present in the images. Finally a performance evaluation is carried out in terms of accuracy based on a comparison of number of blast cells detected by manual count and those detected by this automated method.