White Blood Cell Classification Using Neural Network Approach
| dc.contributor.advisor | Getachew, Alemu (PhD) | |
| dc.contributor.author | Kindalem, Sirak | |
| dc.date.accessioned | 2020-03-06T07:04:14Z | |
| dc.date.accessioned | 2023-11-04T15:14:39Z | |
| dc.date.available | 2020-03-06T07:04:14Z | |
| dc.date.available | 2023-11-04T15:14:39Z | |
| dc.date.issued | 2019-10 | |
| dc.description.abstract | In medical diagnosis blood test is very essential. For this purpose identifying the white blood cell type and recognizing their number are important and useful measure, which indicates the health status of the body. For the analysis of blood cell, laboratory technicians use manual microscopic evaluation which is extremely time-consuming and tedious to segment and classify white blood cells and on the other side, the instruments which are being utilized by specialists for segmentation and classification of blood cells are not economical and affordable for every doctor or hospital. To overcome this, various computational techniques have been developed for segmentation and classification in recent years with improvements in outcomes. In this respect, Artificial Neural Network (ANN) provides the ability and potentials to make classification. The aim of this research work is to design and implement for the classification of white blood cell types from microscopic images of blood samples. Therefore, this research focused on the tasks including the segmentation process, extract suitable features, design the classifier and classify them into five types using the designed ANN model. The system was experimentally analyzed with microscopic images for the classification of the white blood cell types. To acquire region of interest all of microscopic images were segmented. Subsequently, various feature vectors were extracted from the segmented image. After the extraction of feature vectors the classification of each microscopic image for a particular category at the next step was performed using the designed ANN model. The extracted features were used as an input to the neural network. Three feature sets were used to evaluate and compare the performance of the classifier. Accordingly, the segmentation results show that k-means clustering outperforms Otsu thresholding with an average segmentation accuracy of 91.6% and 88.2% respectively. The designed classifier model also yields a classification accuracy of 93.8% to 96.5% based on extracted features from segmented images. It is understood that this research provides the possibility of increasing the speed to find the results of medical analysis by using ANN especially as the number of blood samples increase. | en_US |
| dc.identifier.uri | http://etd.aau.edu.et/handle/123456789/20926 | |
| dc.language.iso | en_US | en_US |
| dc.publisher | Addis Ababa University | en_US |
| dc.subject | classification | en_US |
| dc.subject | microscopic images | en_US |
| dc.subject | neural network | en_US |
| dc.subject | white blood cells | en_US |
| dc.title | White Blood Cell Classification Using Neural Network Approach | en_US |
| dc.type | Thesis | en_US |