Malarial Parasite Detection and Recognition using Microscopic Images

Malaria has been a serious infectious disease since 18th century. Manual diagnosis of malaria is most widely used method but it is a time consuming process, and it involves the risk of error due to the subjective assessment of the sample. In this paper, an automatic method involving image processing techniques is presented which is capable of detecting and recognizing the infection in the microscopic images. The images are generated by using a microscope with 800× zooming capacity. Giemsa staining is used before acquiring the images. Five different categories of malarial parasites are defined and used for classification. Image processing techniques are employed to initially detect a parasite and later to classify it as one of the target categories. Images belonging to three categories were classified perfectly, while one of the category received lower recognition rate. As a result, the proposed method produced 100% classification accuracy for four classes, and 60% for the remaining class. The algorithm developed for classification in hierarchical manner showed good results overall, considering the fact that no such research is available for local data.