Automated Blood Counter (ABC): an Image Processing Solution

Abstract

Nowadays there are many types of research are going on the computer vision area. And there are some new trends for the Bio-medical area with computer vision. But for blood counting purpose there are still we can use image processing solutions for solving problems. To diagnosis of a huge variety of diseases usually, we tend to take a blood sample and reckoning and analysis of blood cells thereon sample. To calculate platelets or blood count, there are two techniques normally use namely, manual technique and an automatic machine technique. The manual enumeration of white blood cells, red blood cells and platelets in microscopic read is an extremely tedious, time intense, and inaccurate method and the count depends on the experience of the lab technicians. In the other hand, automated machines are very fast and highly accurate but, the main problem of this machine is the incredible price and also there are few numbers of units are obtainable in Sri-Lanka and most of these doesn?t seem to be operate properly. Our solution can apply for the manual testing method to reduce the time delay (for counting) and will increase the accuracy of the ultimate output. In the manual process, using a microscope and searching the blood cells through that microscope and do the blood counting process manually. Basically, when we upload those microscopic images to the proposed system, then system will produce the cell count. In our system, we developed algorithm to produce the red cell count based on the image processing techniques. Algorithm used masking and edge detection techniques for identify the cells and for counting purpose it applied contour detection and ellipse fitting method. Challenging task in this research is splitting touching cells. Our proposed algorithm can identify touching cells and using contour detection and ellipse fitting algorithm can segment those touching cells.