Comparative Analysis of Support Vector Machine and Convolutional Neural Network for Malaria Parasite Classification and Feature Extraction

Abstract

Malaria is a dangerous infectious disease because, if it is slow to handle, it can even cause death. Malaria is caused by a parasite called plasmodium, which is transmitted through the bite of a malaria mosquito called Anopheles. Parasites transmitted by mosquitoes attack human blood cells. The inspection method used to identify the type of malaria parasite is microscopic examination, whose accuracy and efficiency depend on human expertise. Examination methods using the Rapid Diagnostic Test (RDT) and Polymerase Chain Reaction (PCR) are not affordable, especially in underprivileged areas. This study compares the performance of classification methods, namely Support Vector Machine (SVM) and Convolutional Neural Network (CNN), to identify the type of malaria parasite and its stage and develop a feature extraction algorithm. The method of feature extraction is a decisive step to identifying the type of malaria parasite. The feature extraction process by developing a feature extraction algorithm is called the PEMA and KEHE feature tracking algorithm, or feature tracking with perimeter, eccentricity, metric, area, contrast, energy, homogeneity, and entropy. The classifier uses a convolutional neural network (CNN) to divide the samples into 16 classes. The experiment used 446 images of malaria parasites. The outcome of identification showed that by tracking the PEMA and KEHE features with the SVM classifier, the best accuracy value was 85.08%, compared to CNN with an accuracy value of 61.40%.