Design and implementation of a computer- based information system by utilizing a developed methodology is presented in this thesis. This methodology includes a set of procedures and processes to analyze blood smear images (image reading, preprocessing, feature extraction, RBCs classification, results and diagnosis).
Essentially an efficient method using the image processing technique (image enhancement, segmentation and feature extraction) has been constructed which can be used to analyze blood smear images taken by photomicroscope. Usually blood smear images contain red blood cells, white blood cells and platelets. In this work the method isolates and determines the type of all red blood cells in the smear which could be either normal or abnormal.
Taking into consideration that abnormal red blood cells indicate to the associated blood anemia. Focus was directed to the process of classifying types of abnormal cells. The system counts the overall red blood cells and calculates percentage of different types of counted abnormal cells (macrocyte, target cell, howel-jolley body, sickle cell, elliptocyte, tear drop, spherocyte, stomatocyte, basophilic stippling, reticulocyte, microcyte, and nucleated RBCs).
The existence of different abnormal types and the related percentages indicate to the type of blood anemia. As a result and to testing the accuracy of these methods (the traditional method and this system) were applied to the same blood samples which contain 597 cells. Results of the two methods were close enough and acceptably comparable. The accuracy of this method is 83%.
Source: MEU
Author: Bashar Abdallah Issa Khawaldeh
>> Design, Modeling and Simulation Projects