Automated status classification of malaria plasmodia from thin blood smears microphotograph using morpho-geometrical feature extraction
Malarial infection analyzed by manually examining a thin blood smear, done by an expert microscopist. Unfortunately, manual examination of the blood slide can be time consuming and is prone to human errors. Hence, this research aims to develop an algorithm that is based on the a priori knowledge of the experts so that the microscopic images can be analyzed with minimal human intervention. In this research, a morpho-geometrical approach of feature extraction coupled with Naive Bayes classification theory are proposed to measure the infected cell’s size and shape to do species and life stage differentiation of P. falciparum, P. malariae, P. ovale, and P. vivax. This is donewith the help of computational geometry, recursive bottleneck detection algorithm, and thresholding with Otsu’s method. In the end, the proposed algorithm when evaluated using real malaria cases produces a PPV (Positive Predictive Value) score of 77.14%, sensitivity score of 84.37%, and an F_1 score of 80.60% which shows that the proposed features are reliable and faithful to the expert’s a priori knowledge.
B02505 | (Rack Thesis) | Available |
No other version available