Recognition of malaria parasites using images of red blood cells

Abstract EN

Malaria is a serious disease in the world and may lead to death if not treated.

It is an infection caused by a single-celled parasite that penetrates the bloodstream through the bite of mosquitoes.

Malaria is diagnosed in several ways, including direct detection by a doctor and microscopic diagnosis by examining blood smears from red blood cells infected with parasites, in addition to the rapid diagnostic test, and these methods are ineffective because of the difference the accuracy of the diagnosis and also its low results in the diagnosis, so it was necessary to use modern technologies to recognize malaria effectively.

In this paper, malaria was recognized by classifying images of infected and uninfected blood cells using the finetuning a pre-trained Convolutional Neural Network as a model for artificial intelligence.

The results of the proposed model were compared with the method of microscopy and rapid diagnosis, and the experimental results showed that the proposed model for the identification of malaria diseases achieved high accuracy and efficiency in terms of performance measures: Accuracy, Sensitivity, Specificity, Precision, F1 score, and Matthews correlation coefficient, where the results were (98.30%, 96.99%, 97.75%, 97.73%, 97.36%, and 94.75%) respectively.