An artificial intelligence algorithm has shown promise in helping monitor and, therefore, stop the spread of facial cancer in wild Tasmanian devil populations.
Devil facial tumour disease (DFTD) is a transmittable parasitic cancer that leads to lesions and lumps forming in and around the mouth of Tasmanian devils. These quickly develop into large tumours on the face and neck, sometimes spreading to other parts of the body.
Devils infected with the aggressive non-viral tumours struggle to feed, causing many infected animals to starve to death. Once the cancer becomes visible, it almost always leads to death.
According to the Menzies Institute for Medical Research at the University of Tasmania, the wild Tasmanian devil population has declined by up to 50 percent since DFTD emerged. Devil populations in roughly 65 percent of Tasmania are believed to be infected, with the highest concentrations of the disease appearing in the island state’s east.
Scientists have been grappling with how to solve the massive impact of DFTD on the Tasmanian devil population since the 1990s when it emerged.
The disease is clonally transmissible – that is, it spreads through the transfer of cells between individual animals. It is most commonly passed on when devils engage in stoushes over prey, territory or mating rights, when they bite each other.
A factor in the spread of DFTD may be the relatively low genetic diversity among Tasmanian devils which became extinct on the Australian mainland about 3,000 years ago and are now isolated in the wild to Tasmania.
One key in fighting to halt the disease’s spread is monitoring DFTD transmission.
Currently, scientists either trap devils or use mounted motion-sensitive cameras to capture images of the animals.
While using cameras allows for more frequent monitoring, differentiating tumours from wounds is challenging and requires time and expertise.
An team of researchers from Turkey’s Near East University and the Tasmanian state government’s Save the Tasmanian Devil Program collaborated to test the success of an artificial intelligence (AI) software tool to determine whether a Tasmanian devil is infected with DFTD.
The scientists used a series of machine learning tools in analysing the pictures. A U-Net neural network architecture segmented the image. Pre-trained Resnet-18 deep learning software then extracted its features. Finally, support vector machine classifiers were used to classify features as tumours or other features.
This setup was tested on 1,250 images of 961 healthy and 289 diseased devils.
They found that the system of AI tools achieved 92.4 percent accuracy in determining whether a Tasmanian devil is infected or not. They write: “The proposed approach will allow for more frequent analysis of devils while reducing the workload of field staff. Ultimately, this automation could be expanded to other species for simultaneous monitoring at shorter intervals to facilitate broadened ecological assessments.”
The research is published in the CSIRO journal Wildlife Research.
Do you care about the oceans? Are you interested in scientific developments that affect them? Then our new email newsletter Ultramarine, launching soon, is for you. Click here to become an inaugural subscriber.