Quantitation of Malarial parasitemia in Giemsa stained thin blood smears using Six Sigma threshold as preprocessor

Malaria is a precarious disease and a serious illness that can become life threatening very fast and can be incurable if not treated on time. According to the World Health Organization, Malaria causes 150-250 million infections and an estimated more than a million deaths each year. Microscopic image capture followed by visual observation as per the manual is the gold standard method for diagnosis of malaria. However, due to subjectivity and the complexity of manual assessment, microscopic diagnosis of malaria is tiring, time consuming and subject to human error. In this study, we have developed a novel automatic high performance method with minimal human dependence for detection and quantitation of malaria infected Red Blood Cells (RBC). The parasites detection test to identify malaria was computed using digital image processing techniques. This automated quantitation procedure was done in three parts: Segmentation, Identification and Detection. The methods used for the quantitation were Six Sigma threshold for segmenting Region of Interest followed by modified Hough transform to identify and count RBCs, and Kapur's threshold method to detect malaria parasite infected RBCs. The developed package using Microsoft® VB.NET 2008 framework is executed on various Giemsa stained thin blood smears digitally acquired using a charge coupled device attached to a microscope. Following analysis performed on seven sets of thin blood smear images, the values of Precision, Recall and F-measures obtained were 96%, 97% and 97% respectively. This is the first attempt to use a combination of Six Sigma threshold, Chess-Board distance, Hough transform and Kapur's threshold to find the RBCs by deriving information from the image itself. Further, Kapur's entropy measure was successfully applied to distinguish parasitized from un-infected cells.