Thesis

Blood Smear Cell Detection and Counting using a Novel K-mean Clustering Algorithm

The segmentation of blood cell is a fundamental and a critical invention in facilitation of automated differential blood cell counting and their subsequent classification and analysis required for clinical examination. Blood cell segmentation is considered to be an essential issue in the hematological studies that lead to diagnosis of diseases like leukemia, anemia and all the other hematological diseases. This paper presents a novel approach to the segmentation of leukocytes and erythrocytes and their subsequent counting in the blood smear image. Our algorithm relies on the K-means clustering technique to attain the segmentation of cells, a few morphological operations to render the image ready for counting using the Circular Hough Transform. The algorithm was tested on 15 different images obtained from LISC database of different color tonality as well different resolution. The algorithm was found to produce a consistent result with an average accuracy of 98.07% in the counting of erythrocytes and segmentation of leukocytes regardless of their resolution.

Relationships

Items