Anaemia cells detection based on shape signature using neural networks

Anaemia becomes more famous and public disease in our life. Many methods have been examined on red blood cells to detect this disease. This paper has been presented an algorithm for detecting anaemia kinds; such as sickle and elliptocytosis dependent on their geometrical shape signatures method. The proposed algorithm presents Circular Hough Transforms, watershed segmentation, and morphological mathematics functions as effective methods to detect the normal blood cells; but the anaemia kinds have been classified based on their shape signatures. Some difficulties have been faced through the detection process; the adhesion of cells may be not belonging to one of previous mentioned kinds of anaemia then they considered cells with unknown shapes. The results of the proposed algorithm have been achieved high precision in all processes steps on experimented 30 colourful microscopic images and the rates have been achieved to 100% and 100% based on three neural networks for all anaemia kinds.