Malaria Detection Using Blood Smear Images

Abstract

Malaria is one of the deadliest diseases. A
perilous disease provoked by parasites that can
disseminated to people from the nips of infected female
mosquitoes which belongs to Anopheles genus. It is
preventable and curable. However, Malaria interpretation
entail close scrutiny of the blood smear images at 100x
enlargement. This is followed by a freehand computing
process in which adepts tally the count of Red blood cells
influenced by parasites. Perception of images of malaria
blood smear is a upgradeable self-activating quick fix
which extricate a sea of time for medical sector plod along
struggle odds with this pernicious disease. In this work, we
set out to recognize from images of blood smear using deep
learning methods to predict in case the sample is taken
from healthy person or not. Here, we use SVM_HOG a
deep learning technique to classify images in
Parasitized/Uninfected images in which we use almost
19290 images of cells where it contain similar amount of
both infected and uninfected images from the Kaggle
database ,from image we extracted the HOG Features after
extracting features we feed to a classifier SVM to predict
whether itis a Parasitized/Uninfected images the accuracy
of our model using the data set of malaria blood smear
images, we attained an accuracy of 92.69% usingLinear
SVM as a classifier . The results suggest that it has high
accuracy on comparision with other techniques.