Splitting of overlapping cells in peripheral blood smear images by concavity analysis

Abstract

The diagnosis of a patient's pathological condition, through the study of peripheral blood smear images, is a highly complicated process, the results of which require high levels of precision. In order to analyze the cells in the images individually, the cells can be segmented using appropriate automated segmentation techniques, thereby avoiding the cumbersome and error-prone existing manual methods. A marker controlled watershed transform, which was used in the previous study is an efficient technique to segment the cells and split overlapping cells in the image. However this technique fails to split the overlapping cells that do not have higher gradient values in the overlapping area. The proposed work aims to analyze the concavity of the overlapping cells and split the clumped Red Blood Cells (RBCs), as RBC segmentation is vital in diagnosing various pathological disorders and life-threatening diseases such as malaria. Splitting is done based on the number of dip points in the overlapping region using developed splitting algorithms. Successful splitting of overlapped RBCs help the count of the RBC's remain accurate during the search for possible pathological infections and disorders. © 2014 Springer International Publishing.