MALARIA PARASITE DETECTION USING DIFFERENT MACHINE LEARNING CLASSIFIER

In the tropical and the subtropical countries, malaria has been a challenge, which really needs a quick and precise diagnosis to put a stop or control the disease. The conventional microscopy method has some shortcomings which includes time consumption and reproducibility. Many of the alternative methods are expensive and it's not readily accessible to the developing countries that need them. In this paper a fast and precise system was developed using stained blood smear images. We employed watershed segmentation technique to acquire plasmodium infected and non-infected erythrocytes and relevant feature was extracted. Six different machine learning techniques for classification are used in the experiments. Fine Gaussian SVM had a True Positive Rate (TPR) of 99.8% in the detection of the plasmodium infected erythrocyte.