Abstract
Background: Malarial fever disease mainly caused by Plasmodium para-site that is infectious to red blood cells. Manual mode of blood cell counting is a tedious process, this leads to distressing method for diagnosis. This process’s mainly impacted on larger screening process.Introduction: The advanced stage of technology, computer aided detection and analysis of this malarial disease, based on Gabor Filters followed by the comparison of XG-Boost classifier, Support Vector Machine and Neural Net-work Classifier algorithms chosen as architecture of choice for recognition and classification of these malarial blood cells.Objective: The goal of this paper is to slow down the complexity in model discrepancy’s, and bring it to more desirable robustness and generalization, through the model development which detects and classify the parasitized and uninfected blood cells in the given sample. Roughly 13750 parasitized and 13750 unparasitized samples was taken for experiments.Results: From the experiments the models such as S.V.M achieved 94% and XG-Boost achieved 90% neural network classifier achieved 80% , out of these S.V.M performed good results in classifying and recognizing the parasitized and uninfected blood cells to increase the accuracy in decision making. Conclusion: The accomplishment of these M.L models, pretends these problems with less variance and obtained excellent results.
A Novel Stacked CNN for Malarial Parasite Detection in Thin Blood Smear Images
