Performance of Convolutional Neural Network in Detecting Plasmodium Parasites
Hanung Adi Nugroho, Julisa Bana Abraham, Aina Hubby Azzira, Eka Legya Frannita, Rizki Nurfauzi, Faza Maula Azif, E. Elsa Herdiana Murhandarwati · 2019
Automated Plasmodium detection still remains a challenging problem in biomedical engineering field. The variation of data becomes one of the problems that should be carefully handled. Whilst, deep learning approach is growing rapidly. Convolutional neural network (CNN) is one of the deep learning methods that is able to learn the characteristic of object effectively and had been widely used in several fields including biomedical engineering. Hence, this research work proposes a scheme to detect Plasmodium on thin blood smear images using CNN model. In this work, two CNN architectures are compared. The two architectures are simple CNN with two convolutional layers and transfer learning model with pretrained ResNet-50 as the feature extractor. The transfer learning model successfully achieves positive predictive value and sensitivity of 98.46% and 97.00%, respectively. This result indicates that the proposed scheme is suitable for detection of Plasmodium parasites.