GEOMETRIC FEATURES: A CRITICAL COMPONENT FOR ACCURATE MALARIA PARASITE STAGE CLASSIFICATION IN THIN SMEAR MICROSCOPY
Gunjan Aggarwal, Mayank Kumar Goyal · ShodhKosh Journal of Visual and Performing Arts · 2024
Malaria-a global health problem always demands an accurate and timely diagnosis of a disease for its proper treatment. Traditional methods like microscopic examination are time-consuming and require specialized expertise. It thus poses challenges in resource-limited areas. Automated classification of malaria parasite stages helps in improving the diagnostic efficiency. In this paper, the importance of geometric features in malaria parasite stage classification using machine learning techniques has been realized. Geometric features, including area, perimeter, and shape descriptors, offer valuable information regarding the morphological differences between the various stages of the parasite. We compare the performance of the following machine learning models using geometric features: Random Forest, GaussianNB, XGBoost, and MLPClassifier. The results show that the inclusion of geometric features improves the accuracy and robustness of the machine learning models for classification. Among the different models tested in this study, MLP Classifier had 95.90% accuracy thus shows tremendous potential for a geometric feature in a malaria diagnosis program. This current study, therefore, gives way to advancement in automated diagnosis of malaria among others and further pursuit of geometric-based applications in their fields.