Automated Malaria Parasite Detection Using Artificial Neural Network

Malaria is still an infectious disease that causes high mortality in endemic regions. It is thought that it will maintain importance in the future, especially due to people travelling from African countries where malaria is endemic to its eradicated regions. Therefore, rapid and accurate diagnosis is a critical step in the effective treatment of malaria and reducing mortality rates. This paper provides a malaria diagnosis system using an artificial neural network approach with SURF (Speeded Up Robust Features) method that helps the clinicians to predict and locate infected cell with malaria on the sample thin blood smear image. The performance of the proposed neural network and local image feature extraction technique SURF were analyzed statistically and presented in this paper. The network was trained using only 45 infected thin blood smear images and was then tested with 200 (100 infected and 100 non-infected) unseen images. The experimental results showed that the proposed system identified the malaria parasite with 93% accuracy, 86% sensitivity and 100% specificity.