Please use this identifier to cite or link to this item:
|Title||Detecting Red Blood Cells Morphological Abnormalities Using Genetic Algorithm and Kmeans|
Vision is the most advanced of our senses, so it is not surprising that images play the single most important role in human perception. Computer-aided diagnosis is another important application of pattern recognition, aiming at assisting doctors in making diagnostic decisions. Many diseases which are not blood diseases in origin have hematological abnormalities and manifestation (have symptoms appeared on the blood). CBC (cell blood count test) for instance, is still the first test to be requested by the physicians or become in their mind. Blood abnormality can be in white blood cells, red blood cells and plasma. In this thesis, red blood cells are the suggested for detecting it is abnormality. The abnormality of blood cells shapes can't be detected easily, where the CBC (cell blood count) device give a count number and percentages not a description of the shapes of the blood cells, when the blood cells shapes wanted to be known, hematologist asked to view the blood films under the microscope which is time consuming task besides that the human error risk is high. Since the number of abnormal cells to normal cells in a given blood sample give a measure of the disease severity, detecting one cell with potential abnormality can give premature warning for future illness that can be avoided or treated earlier. This case can't be detected by hematologist. Computer involved in such task to save time and effort besides minimizing human error. This thesis name is "DETECTING RED BLOOD CELLS MORPHOLOGICAL ABNORMALITIES USING GENETIC ALGORITHM AND KMEANS". In this thesis, the thesis divided into four phases. First phase data collection where blood samples was drawn from healthy and sick people and then blood films made and viewed under microscope and an images captured for these blood films. Second phase preprocessing phase where the images prepared for the next phase. Third phase feature extraction was executed where these features are spatial domain and frequency domain features. Fourth phase is the classification phase where the features fed into the classifier to be classified. An acceptable detection rate is achieved by the proposed system. The genetic algorithm classifier success rate was 92.31% and the K-means classifier success rate was 94.00%.
|Publisher||the islamic university|
|Files in this item|