An automated method to the identification and counting of blood leukocytes and thrombocytes using fuzzy clustering and support vector machine from microscopic blood images
An automated method to the identification and counting of blood leukocytes and thrombocytes using fuzzy clustering and support vector machine from microscopic blood images
Sima Amin ol Eslami1 Siamak Haghipour2
1) Iran, Islamic Azad University, Tabriz Branch /
2) Iran, Islamic Azad University, Tabriz Branch /
IOI Code
:
XYCZ-AAHZF
Publication :
International Conference on Engineering and Applied Science(iceasconf.com)
Abstract :
Counting and classifying blood cells provide an invaluable information for the pathologist to diagnose and treat illnesses. As the manual identification and counting of blood is tedious, time-consuming and relatively inaccurate, presented an identification system for auto-analysis of blood cell images based on image processing and computational intelligence methods will be effective. In this study, an automated method to the identification and counting of blood leukocytes and thrombocytes is presented. With new image processing methods taken into account, hough transfoms method was chosen for the identification of overlapping cells, new algorithms were developed based on fuzzy data mining to improve the flaws of hough transform followed by the proposition of the support vector machine classifier as a means to optimize the flaws of colorful filters. In addition, two image groups were used for the classification. first group for training and the second group for testing. In fact, using fuzzy clustering, desirable radii were extracted automatically. Subsequently, the location of the overlapping cells was found by hough transform as well as being classified by colorful support vector machine. In training phase, results indicate an accuracy of 100% and 83.87% respectively in the identification of leukocytes and minute thrombocytes. In testing phase, results indicate an accuracy of 95.55% and 73.60% respectively in the identification of leukocytes and thrombocytes.
Keywords : counting leukocytes thrombocytes fuzzy clustering support vector machine