Segmentation and Evaluation of White Blood Cells using Segmentation Algorithms

Image Segmentation is an important step in Image processing applications. Microscopic images of White Blood Cells (WBC) enable hematologists to predict the vulnerability to several diseases. Automatic segmentation of different types of WBCs is the most challenging task in the prediction of the presence of disease. Our objective is to segment the blood cells in microscopic images using segmentation algorithms followed by evaluation of their performance. Color is a major cue to discern segmented WBCs from microscopic images. This paper appraises the performance of various color based segmentation techniques, and compare the results against ground truth image. Analysis of the results using dice similarity, shows that saliency based segmentation is the most suitable model to segment WBC cells.