Circlet Based Framework for Red Blood Cells Segmentation and Counting

Abstract

The number of Red Blood Cells (RBCs) from blood smear is very important to detect as well as to follow the treatment of many diseases like anemia and leukemia. The old conventional method of RBC counting under microscope gives an unreliable and inaccurate result depending on clinical laboratory technician skills. So, automation of counting is helpful for improving the hematological procedure and reducing time and labor costs. This paper introduces a novel method for RBCs segmentation and counting from microscopic images using Circlet Transform which operates directly on gray-scale image and does not need further binary segmentation. First, mask of RBCs is obtained. Next, circlet transform is applied on gray-scale image. Then, minimum and maximum number of RBCs is estimated. Finally, RBCs are detected and counted by using an iterative soft-thresholding method and removing conflict RBCs. The proposed method outperforms other methods in terms of accuracy.