Detection of some anaemia types in human blood smears using neural networks

The identification process based on measuring the level of haemoglobin and the classification of red blood cells using microscopic examination of blood smears is the principal way to diagnose anaemia. This paper presents a proposed algorithm for detecting some anaemia types like sickle and elliptocytosis and trying to count them with healthy ones in human red blood smears based on the circular Hough transform and some morphological tools. Some cells with unknown shapes (not platelets or white cells) also have been detected. The extracted data from the detection process has been analyzed by neural network. The experimental results have demonstrated high accuracy, and the proposed algorithm has achieved the highest detection of around 98.9% out of all the cells in 27 microscopic images. Effectiveness rates up to 100%, 98%, and 99.3% have been achieved by using neural networks for sickle, elliptocytosis and cells with unknown shapes, respectively.