Segmentation of HEp-2 cells centromere pattern

According to NIH, an autoimmune diseases have infected almost 23.5 million peoples of the USA. The detection of such autoimmune disease is done by using Antinuclear Antibodies test of Human Epithelial pattern 2 (HEp-2) cells. It is needed to design an automatic system for ANA testing, which involves the stage of segmentation of HEp-2 cells. This paper describes efficient algorithm for segmentation of Centromere pattern of HEp-2 cells which is a very difficult task. The proposed algorithm is based on thresholding and morphological operations. The performance of this algorithm is evaluated on public dataset of HEp-2 cells. The separate result are calculated on training and testing dataset which outperformed earlier algorithms in the case of Centromere pattern.