Detection of Malaria Parasite in Blood Using Deep Learning

Abstract

Malaria is one of the severely toxic diseases which occurred or began due to the parasitical nature of the genus Plasmodium, causing millions of mortalities. Microscopy done in a typical, ordinary way, which in general is known as “the gold standards” for malaria diagnosis has sporadically verified as incompetent because it is a prolonged method and the outcomes are tedious to reproduce, hence the quick diagnosing is the extremely required as of the era. To strengthen the results of the diagnosis we can effectively use the concept of image processing The methods proceeds in steps which includes image preprocessing, Image segmentation and red blood cells detection in the first step, Different features like Color, texture, etc. are used to distinguish between blood cells and for this feature extraction and selection is done and finally Convolutional neural network is used for classifying the infected and uninfected red blood cells in thin smears of blood. The system allows the user to give the sample image to be tested and it automatically identifies whether the blood sample is infected or not. This system helps in reduction of time used for diagnosis and curtails the possibility of human blunders to some extent.  The system provides an intelligent and automated way to detect malarial parasites in blood cells. The system makes use of Deep learning concepts and the system uses the Convolutional Neural Network Architecture. We have built our system with the help of a dataset of 27000+ microscopic blood images from various patients.