deleo-lab/schisto-parasite-classification

Image classification for snail and parasite associated with schistosomiasis in Senegal Africa # Schisto-parasite-classification CNNs, transfer learning, VGG16 pre-trained model, Keras, TensorFlow backened ## Introduction Schistosomiasis is a neglected tropical disease (NTD) infecting over 250 million people worldwide. The current approach to mitigate this disease is to deliver the drugs to treat needed communities. However, parasites are primarily transmitted through environment reservoirs where freshwater snails serve as intermediate hosts. People use the contaminated water source for their daily tasks and get re-infected after the drug treatment. Therefore, drug administration alone is not effective for Schistosomiasis control. Recent studies [1, 2] show that snail population control is essential to spot disease transmission risks. To discerning between human parasitic worms and other non-human parasitic species in snails is a necessary step to precise quantification of human risk. As part of Stanford's Program for Disease Ecology, Health and the Environment, our team has collected 5,543 images of freshwater snails as well as 5,140 images of parasitic cercariae, from 11 morphospecies liberated from the Biomphalaria and Bulinus spp. snails encountered and dissected, in Senegal Africa during 2015-2018. This dataset will be made public for the research community (download here). Our research team have used these images as training dataset to develop a deep learning model that classifies images of 11 parasite categories with higher accuracy (88%) than well-trained human parasitologists. Details of this research can be found in the publication [3]. Our model facilitats transfer learning using VGG16 and convolutional neural networks (CNNs). In addition, by working with IBM's Cognitive Open Technologies & Performance Group, we have built a Javascript web application that incorporates with our classification model, which allows users to select their own parasite images and the classification model will provide prediction suggestions for your parasite images. Th …