A comparison of plasmodium falciparum identification from digitalization microscopic thick blood film

摘要

Malaria is a serious public health problem in Indonesia. Conventional methods of identification malaria parasite are generally carried out by paramedics when they are thoroughly examine blood performed using a microscope. This way is currently used anywhere, because it is cheap and it has good accuracy than others. However, this conventional methods can make a difference if the diagnosis is made by different experts. The detection of malaria parasite is time consuming and subjective factors are very high. Therefore, required the appropriate method to identify malaria parasite with a high accuracy. This research aims to compare the level of accuracy among several methods that are used to classify plasmodium falciparum. The comparing methods are KNN (K-Nearest Neighbor), backpropagation, and LVQ (Learning Vector Quantization). This research has three main stages, they are preprocessing, feature extraction, and classification. The preprocessing aims to get ROI (Region of Interest) by cropping manually and resizing images. The feature extraction method uses Gray Level Co-occurrence Matrix (GLCM) to get texture feature values such as contrasts, correlations, energys, and homogeneity that appearance in digitalization microscopic thick blood film. The classification is doing experiments for three classification methods and comparing each method with its accuracy value. The result of comparison algorithm is KNN (K-Nearest Neighbor) has highest accuracy value with recognition rate 84.6667%.