期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2021
卷号：48
期号：4
语种：English
出版社：IAENG - International Association of Engineers
摘要：Detection of white blood cells on a microscope slide using image processing technology has been widely used in health-related research for over a decade. However, most of the proposed research is performed on blood cells with colorstained images. These images are obtained by capturing microscope slide images that have been previously stained with various substances. Unfortunately, this task is very rigorous and time—and—cost inefficient. This study attempted to eliminate the staining phase in the preprocessing stage and aimed to develop an automatic system to detect white blood cells from unstained microscopic images. A modification of the watershed segmentation method was applied to the developed system to differentiate white blood cells from other cells. First, a preprocessing stage which applies various image enhancement techniques was performed to improve the image quality, and the red blood cell areas were subsequently eliminated using the modified watershed segmentation method. The final result was determined by the area size and aspect ratio of the segmented area. Experimental results showed that the average detection accuracy for locating white blood cells from unstained images was 65.42%, while 94.82% average detection accuracy was achieved for color-stained images. Several experiments using the same datasets were also executed by applying the Convolutional Neural Network (CNN) method. The obtained average detection accuracies for the CNN method were 65.33% and 74.85% for the unstained images and the color-stained images, respectively. The comparison of the two methods showed that the developed system can effectively perform blood cells detection without using the color-stained images.
