Malaria is a hazardous disease responsible for nearly 400 to 1000 deaths annually in India. The conventional technique to diagnose malaria is through microscopy. It takes a few hours by for an expert to examine and diagnose malarial parasites in the blood smear. The diagnosis report may vary when the blood smears are analyzed by different experts. In proposed work, an image processing based robust algorithm is designed to diagnose malarial parasites with minimal intervention of an expert. Initially, the images are enhanced by extracting hue, saturation and intensity planes followed by histogram equalization. After preprocessing, a median filter is employed to eliminate the noise from the images. After the preprocessing, segmentation of malaria parasite is achieved using k-means clustering to get the clear vision of the region of interest. The clustering is followed by region growing area extraction to remove the unwanted area from the segmented image. The second part deals with counting the number of infected RBCs. The method uses roundness detection for the calculation of the infected RBCs. The experiments give encouraging results for the saturation plane of HSI color space and segmentation accuracy by up to 94%.
CITATION STYLE
Bhattacharya, S., Bhan, A., & Goyal, A. (2020). Hybrid Segmentation of Malaria-Infected Cells in Thin Blood Slide Images. In Advances in Intelligent Systems and Computing (Vol. 941, pp. 925–934). Springer Verlag. https://doi.org/10.1007/978-3-030-16660-1_90
Mendeley helps you to discover research relevant for your work.