Application of Machine Intelligence in Digital Pathology: Identification of Falciparum Malaria in Thin Blood Smear Image

Advancement of microscopic imaging has contributed to the development of digitized blood smear images. Application of automated algorithms to analyse these images can help pathologist for a rapid and accurate diagnosis. This chapter proposes a new algorithm that uses a combination of both supervised and unsupervised learning methods as well as rule-based methods, for faster and effective segmentation of parasite-infected cells. Morphological and textural features are extracted for identification of parasite stage using publicly available MaMic dataset. The infected cells are segmented with very high accuracy for better classification results. Single classifiers and ensemble of classifiers were used for best classification outcome. The best overall image-level accuracy of 98.74% was achieved by ensemble classifier.