Malaria, a mosquito-borne infectious disease affecting humans and other animals, is widespread in the tropical and subtropical regions. Microscopy is the most common method in diagnosing the malaria parasite from stained blood smears. However, this procedure is time-consuming, error-prone, and requires a well-trained professional. Moreover, the recognition of a malaria parasite through a microscope is still a challenging process, especially in distinguishing multiple stages of parasites.
In this paper, we demonstrate a novel deep learning approach for multi-stage malaria parasites recognition in blood smear images using Deep Transfer Graph Convolutional Network (DTGCN). In addition, this paper is the first application of Graph Convolutional Network (GCN) on multi-stage malaria parasite recognition in blood smear images. The proposed DTGCN model is based on unsupervised learning by transferring knowledge learned from source images that contain the discriminative morphology characteristics of multistage malaria parasites. This transferred information guarantees the effectiveness of the target parasite recognition. This approach firstly learns the identical representations from the source to establish the topological correlations between source class centers and the unlabelled target samples. In this stage, the GCN is implemented to extract graph feature representations for the multistage malaria parasites recognition. The proposed method showed higher accuracy and effectiveness in publicly available microscopy images of multi-stage malaria parasites compared to a wide range of state-of-the-art approaches. Furthermore, this method is also evaluated on a large-scale dataset of unseen malaria parasites and the Babesia dataset. Code and dataset are available in GitHub under an OSI compliant license (MIT).

Keywords:

• malaria
• multi-stage recognition
• microscopic image analysis
• knowledge transfer
• graph convolutional network
• deep learning

Read the peer-reviewed publication(s):

(PubMed: 34137821)

Additional information:

https://scicrunch.org/resolver/RRID:SCR_020976

Github links:

https://github.com/senli2018/DTGCN_2021