Hybrid Approach towards Malaria Parasites Detection from Thin Blood Smear Image

Malaria is still reckoned as a killer disease particularly in tropical countries. This work is aimed at proposing a hybrid approach for malarial parasite detection within thin blood smear images at any stage of infection, ranging from the initial to the final stage. Rule-based methods based on malarial parasite morphology had been proposed for detection of malarial parasite from digital images. However, as a predictable consequence of constant evolution, rule-based methods were unable to scale up to the constant change that is characteristic to evolution. Over the last 10 years, machine learning methodologies which are largely dependent on dynamic rules have been extensively used for malarial parasite detection from microscopic image/s. However, the research conducted over the last 10 years can by far be extended and the efficiency of the prediction system can be enhanced by using a hybrid approach, an approach that uses the best of the past in amalgamation with the new technology at hand. In comparison to other state-of-the-art methods on the same publicly available dataset, the proposed algorithm achieved a Sensitivity and Specificity value of 0.984 and 0.976 respectively.