Malaria is a severe infectious disease caused by a peripheral blood parasite of the genus Plasmodium. In this work, a proposed approach primarily focuses on image processing techniques to process and enhance stained thin blood smear images for feature extraction, as well as machine learning techniques for the final classification of feature space. In the past, conventional microscopy techniques have proven to be time-consuming and had observed a lack of differentiation due to poor accuracy and lack of algorithms used. Researchers in this domain have already used various preprocessing, segmentation, and feature extraction techniques. In this project, our emphasis is to address the issues of conventional microscopy methods using techniques such as Otsu’s method and Watershed algorithm for segmentation, followed by extracting texture features using CNN. We have also calculated color and texture features which also equally play a pivotal role in feature extraction to perform Bins Approach for classification.
Dataset used: The proposed algorithms have been experimented using the subset of Lister Hill National Center for Biomedical Communication (LHNCBC) dataset, which is a part of the National Library of Medicine (NLM). Thanks to the National Institute of Health (NIH) the malaria dataset is available for download from https://ceb.nlm.nih.gov/proj/malaria/cell_images.zip
The performance of the algorithms is evaluated and compared using different performance evaluation parameters like accuracy, precision, and recall. It is expected to obtain better results of classification concerning these parameters.