An automated method for counting and characterizing red blood cells using mathematical morphology

This paper presents an automated method for counting red blood cells present in a blood sample. The proposed method addresses the problems of holes present in blood cells and overlapping characteristics of the red blood cells. The procedure is quite simple and straightforward, which utilizes mathematical morphological operations of erosion and dilation for performing different steps. It first thresholds a gray scale image to obtain the binary image using the Otsu thresholding method, and then, performs the hole filling process on the red blood cells if they have holes. Then, the process moves on to the job of counting the red blood cells. For this, each red blood cell is extracted and its shape analysis is performed to decide whether it is circular, non-circular, overlapping or just partially present in the sample. If a cell is only partially present in the image, then it is discarded. In case of overlapping, the number of cells in the overlapped area is determined. Several experimental results have been presented to establish the effectiveness of the method. One of the important findings is that the proposed method gives accurate count of red blood cells of the blood sample, and classifies each cell into one of the four categories mentioned above.