We collaborate with the team of Prof. Karl Hoffmann from Aberystwyth University to refine image analysis phenotyping of whole organisms for anthelmintic screening and to provide informatics support for their Roboworm platform.
As a part of this engagement, we have created predictive models based on the phenotype of the larval newly excysted juvenile (NEJ) stage of the liver fluke, Fasciola hepatica. F. hepatica is responsible for causing the foodborne zoonotic disease, fascioliasis, affecting 2.4 million people globally, and is estimated to result in the loss of livestock equating to >£645M/annum globally.
Biomedical Image Gateway (BIG) was used to perform segmentation of individual NEJs and to calculate various histogram-based, texture and morphological descriptors. A subset of those descriptors was next selected to build an SVM model using R. Moreover the segmented larvae were used to retrain a CNN model using TensorFlow. The work also benefited from a novel method of mosaicing images with limited information content implemented in BIG.
The results were presented by Dr Kezia Whatley on a poster at the 2018 British Society for Parasitology Spring Meeting in Aberystwyth, which is available as a PDF here.