- A new model developed by researchers at the University of Waterloo, Ontario in Canada uses spectroscopy and AI (artificial intelligence) technology that can identify the presence of microplastics in wastewater.
- The researchers trained PlasticNet to detect microplastics based on the way they absorb and transmit various wavelengths of light they are exposed to.
- The tool successfully classified 11 types of common plastics with an accuracy of more than 95%; which could potentially be used by wastewater treatment plants and food manufacturers to identify microplastics.
- The team is currently working on making the model work faster and more efficiently, as well as simplifying the data collection process.
The presence of microplastics has become a problem in the aquatic environment. The existence of plastic pieces less than 5mm long has now become an acute global problem. So how do you find out about microplastic pollution in water and sea waters?
This question haunts Wayne Parker, a professor of civil and environmental engineering at the University of Waterloo in Ontario.
“Plastic pollution has caused concern, our aim is to determine the level and quantity of microplastics that are dumped into rivers and lakes. "If anything, the current technology is very time consuming and impractical," explained Parker in an interview with Mongabay.
So far, optical or infrared microscope technology has been used to manually analyze and identify small fragments of plastic waste in the water. However, distinguishing microplastics from other substances takes time and expertise.
“As we started to learn more advanced microscope methods, we found that analyzing the images produced by those microscopes had some problems,” Parker said.
"We then thought about whether this could be solved through deep learning and artificial intelligence (AI)-based approaches . "
Parker then contacted his colleague, Alexander Wong, an artificial intelligence expert and professor in the systems design engineering department at the same university.
Together with PhD candidates they then developed an image identification system that can be used by wastewater treatment plants and the food industry to identify microplastics in the wastewater and food products they produce.
Finally, these researchers succeeded in creating an artificial intelligence-based deep learning program called PlasticNet. This tool works by identifying microplastics based on signal responses from light exposure.
The study published by the team in the journal Environmental Pollution details how they trained a model to detect microplastics based on their interaction response with different wavelengths of light.
Parker's team used advanced spectroscopic techniques in which they illuminated the water. Different types of plastic are capable of absorbing and transmitting specific light at different scanning wavelengths.
Although not yet generally available, this tool has been proven to be able to identify microplastics accurately and quickly. After being ‘trained’ with 8000+ pure plastic spectra, PlasticNet successfully classified 11 common types of plastic with an accuracy higher than 95%.”
As a first step, AI was trained using polymer plastic. Furthermore, plastics made from more complex materials are used, not just pure polymers but containing additives that give color or other chemicals.
"That will certainly have a different impact on light transmission," said Parker.
As plastics became more complex with varying shapes, sizes and thicknesses, the team observed improvements in model accuracy and efficiency. They compared their approach with conventional microplastic approximation methods.
The results, Parker said: "This method is 50% faster and more accurate."
Does all this make the researchers satisfied with their findings?
Parker said this model could still be improved, he said he would train the AI with a variety of more complex plastics in the next few months. The images obtained will contain more detailed and concise information.
“We certainly have to consider whether we can be more efficient in terms of the amount of data we need to try and get, we [are trying] to speed up the overall analysis to be more efficient.”
Original post: New AI model helps detect and identify microplastics in wastewater. This article was translated by Akita Verselita.