DEEP LEARNING APPROACH FOR AUTOMATED SCREENING OF MALARIA PARASITE
Srujana Inturi · Journal of Emerging Technologies and Innovative Research · 2021
Malaria is a life-threatening disease caused by Plasmodium parasites that infect the red blood cells (RBCs).According to the 2018 World Health Organization (WHO) report, an estimated 435,000 malaria- related deaths are reported globally. Children under 5 years of age are reported to be the most vulnerable, accounting for 61% of the estimated death counts. Early diagnosis and treatment is the most effective way to prevent the disease. Manual identification and counting of parasitized cells in microscopic thick/thin-film blood examination remains the common, but burdensome method for disease diagnosis. Its diagnostic accuracy is adversely impacted by inter/intra-observer variability, particularly in large-scale screening under resource-constrained settings.The primary aim of the paper is to overcome these inter/intra-observer variability or errors that occur as a result of these by automating this process using Deep Learning Algorithms like Convolutional Neural Network (CNN) in such a way so as to have the maximum accuracy. These CNN’s can be used to detect whether a Red Blood Cell stained with a chemical Giemsa is infected with Malaria or not. The Malaria Cell Image Dataset provided by Kaggle consisting of a total of 27,558 images of the blood cells is being used to train the model.This data can be used to train various pre trained CNN models like ResNet- 152, WideResNet101-2 as they have low top-1 and top-5 errors. Instead of using the pretrained models as it is, we can unfreeze the already existing layers and modify the layers to suit our dataset. The dataset will also be tested using SGD+Nesterov+Momentum,Adam AdaGrad, AdaMax optimizers and the best combination would be used for the final model.This CNN model would then be either deployed directly in an Android Phone or can be deployed in a remote web server to which images can be sent, processed at the server and results sent back