Malaria Disease Prediction using Faster RCNN

Malaria is a disease caused by Plasmodium parasites that remains a major threat to global health. It affects 200 million people and causes 400,000 deaths each year. The Plasmodium parasite, which is spread through the bites of female Anopheles mosquitoes, is the main cause of malaria. Typically, the use of a microscope in microbiological studies facilitates the identification of infected cells in a blood sample, followed by expert analysis of the results to complete the diagnostic process. This type of expert analysis does not give a completely 100% perfect conclusion and it also costs more. Identifying such targets is also a challenge, largely due to differences in cell shape, density and color, and the ambiguity surrounding certain cell classes. Therefore, a method based on deep learning is used to predict malaria with the highest accuracy. Deep learning-based technologies have proven to be able to achieve human-level accuracy in object detection/classification in image data. These approaches can be used to automate many of the monotonous tasks involved in analyzing micrographs of blood samples. It uses Faster R-CNN. The proposed models were trained and tested on a publicly available erythrocyte image dataset containing both infected and uninfected cells. Methods developed to fine-tune the images using image pre-processing successfully identified malariainfected erythrocytes with the highest accuracy in ashort period of time. The software is also adapted to a budget microcomputer to speed up prototyping.