DETECTION OF MALARIA PARASITES IN HUMANBLOOD CELLS USING CONVOLUTIONAL NEURALNETWORK

Abstract

Malaria is a blood disease caused by the Plasmodium parasite
which is transmitted by the bite of the female Anopheles
mosquito. The diagnosis of malaria is carried out by a
microscopist through examination of human blood cells. Their
level of accuracy depends on the quality of the tool, expertise
in classifying and counting infected and uninfected parasite
cells. The disadvantages of examining this way include the
difficulty in making a diagnosis on a large scale and the poor
quality of the results. The dataset used in model evaluation is
a dataset developed by LHNVBC which contains 27,558 cell
image data. The malaria dataset will be processed through data
science processing using a Convolutional Neural Network
with the ResNet architecture. The model will conduct training
on the dataset and then the model will be able to recognize
malaria parasites in human blood cells. The model will be
trained by optimizing multinomial logistic regression using
Stochastic Gradient Descent (SGD) and Nesterov momentum
values. The results of training data validation accuracy from
model training with 50 epochs were obtained at 96.23% and
97% after being tested on data testing.