Open AccessFeature analysis for stage identification of Plasmodium vivax based on digital microscopic image
Author(s) -
Hanung Adi Nugroho,
I Made Dendi Maysanjaya,
E. Elsa Herdiana Murhandarwati,
Widhia K.Z. Oktoeberza
Publication year - 2019
Publication title -
indonesian journal of electrical engineering and computer science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.241
H-Index - 17
eISSN - 2502-4760
pISSN - 2502-4752
DOI - 10.11591/ijeecs.v13.i2.pp721-728
Subject(s) - parasite hosting , artificial intelligence , pattern recognition (psychology) , plasmodium vivax , malaria , gametocyte , digital image , computer science , feature selection , image processing , naive bayes classifier , computer vision , biology , plasmodium falciparum , image (mathematics) , support vector machine , immunology , world wide web
Plasmodium parasite is identified to confirm malaria disease. Paramedics need to observe the presence of this parasite prepared on thick and thin blood films under microscope. However, false identification still occurs which is caused by human factor during the examination. Thus, malaria identification based on digital image processing has been widely developed to overcome the error possibility. This paper proposes a scheme to identify and classify the stages of Plasmodium vivax parasite on digital microscopic image of thin blood films based on feature analysis. Shape and texture features are extracted from segmented parasite objects. Feature selection based on wrapper method is then conducted to obtain relevant features which may contribute in improving the classification result. The classification process is conducted based on Naïve Bayes classifier. The performance of proposed method is evaluated using 73 digital microscopic images of P.vivax parasite on thin blood films comprising of 29 trophozoites, 10 schizonts and 34 gametocytes stages. By using six selected features including perimeter, dispersion, mean of intensity, ASM, contrast GLCM and entropy GLCM, the proposed scheme achieves the best classification rate with the accuracy, sensitivity and specificity of 97.29%, 97.30% and 97.30%, respectively. This indicates that the proposed scheme has a potential to be implemented in the development of a computerised aided malaria diagnosis system for assisting the paramedics.