Parasites have the most significant impact on animal performance after nutrition. Parasites also have a massive effect on farm productivity; management of parasitism consumes a considerable amount of time and money. The way parasites are diagnosed had not changed for decades until Techion developed FECPAK. Techion’s newest product FECPAKG2 modernises the traditional microscope-based testing method with an image-based, internet-connected, software supported diagnostic platform.
Techion is a Dunedin based company, and our technology has been developed right here in association with Otago University. The technology we use to accumulate eggs for imaging and counting is patented around the world in our major markets.
We have customers worldwide, and have been involved in many projects with international collaborators, including Starworms; a project supported by the Gates Foundation.
This project aims to extend the speed and capability of our diagnostic platform by enabling accurate AI counting and recognition of important species of parasite eggs.
Traditional parasite egg counting in samples is done using microscopes, manual counting, and transcribing of results which is time-consuming, expensive and prone to mistakes. The species of parasite can be significant – some species are very harmful, some are benign. Treatment can vary depending on species. Identification to the species level is currently made via larval culture which is time-consuming. Development of this technology will enable real-time, fast and more helpful results for farmers.
- 2 semesters
You will gain a full understanding of the background of the project from your mentor John Newton who has been conducting most of the research in this area for Techion. Using information you have gained from John and research of your own you will undertake and report on a literature review of AI applications in microscopy and identify potential competitors or collaborators for Techion in this area.
We have an early stage AI model and codebase written by John in FastAI & PyTorch. The model can currently count eggs using Techion’s FECPAKG2 patented well technology with an accuracy of >95%. You will further develop the AI model to distinguish, classify and count eggs from the Nematodirus and Strongyle parasites in well images produced from faecal samples using FECPAKG2. FECPAKG2 is a novel parasite separation automatic microscope platform. Algorithm development will be done using a convolutional neural network (CNN). Basic documentation of all tools and programs developed will be required.
Note: you won’t need to be anywhere near the samples, and all work can be done remotely online.
The images generated using the FECPAKG2 well technology contain multiple species from within the strongyle and Nematodirus genera that cannot be differentiated by the human eye. We have contacts internationally who are working with us to achieve this speciation. For example, scientists at Cornell University have FECPAKG2 and they have been creating images of Haemonchus Contortus eggs and sending them to us. You will work with international collaborators such as these, to create libraries of single species images. You will then use the single species images to develop ways to distinguish and count different strongyle species and different Nematodirus species in the same sample in images from our FECPAKG2 wells.
As well as diagnosing strongyle and Nematodirus parasites the FECPAKG2 system is being developed to identify and count other species of eggs, including, but not limited to, Coccidia, rumen fluke and liver fluke. These eggs are imaged using different technology on fine film slides. You may be developing additional AI models to identify and count these additional species as part of your project.
Techion is seeking suitable students with a relevant undergraduate degree who are interested in gaining experience in practical applications of AI in technology. Familiarity with Jupyter Notebook, PyTorch and python is needed. Techion will provide a subscription to Azure or Google cloud for the duration of the project.
An interest in microscopy would be an advantage but is not essential.
We have an advisor, John Newton also based in Auckland, and you would work closely with him. John has been developing Techion’s new hardware platform for the past two years and developed many of the optical algorithms used on our product. John has a Masters in Science and has completed several deep learning courses.
We expect you will visit us in Dunedin at least once during the year to get first-hand experience of how the technology you develop will be used to help farmers and veterinarians. We rely heavily on Zoom for our face to face meetings with our many colleagues worldwide.
This project qualifies for a Callaghan Innovation R&D Fellowship Grant. The successful candidate will get paid a stipend of $26,000 p.a. (this is a one year project).
For students to be eligible for an R&D Fellowship Grant placement, students must: