Device can detect cancer
Kitchener firm finds all it needs here to develop artificial intelligence product
A hand-held device using the latest technology in artificial intelligence and machine learning will help doctors quickly determine whether new moles and lesions are early-stage melanoma — the deadliest of skin cancers.
A team of researchers from the University of Waterloo working with professor Alexander Wong in systems design engineering, founded Elucid Labs to bring this emerging technology to market.
That could happen in less than a year, said Wong.
Currently, doctors rely on visual examinations and then order biopsies of suspicious looking spots. Biopsies are expensive and invasive and most come back negative for cancer.
“So that results in a lot of high costs in the health-care system, a lot of surgical procedures and complications,” said Wong.
“Our goal was: Can we actually figure out these biomarkers, without touching the patient, through the power of imaging?” he said.
The Elucid scanner detects biological changes in the skin associated with melanoma — the concentration and distribution of eumelanin, the chemical that gives skin its colour; and hemoglobin, a protein in red blood cells.
At the heart of the device is a cutting-edge application of artificial intelligence, or AI.
“What we were able to do is build an AI that is able to identify and extract these quantitative biomarkers that can be used by dermatologists and other clinicians to assess whether this mole or lesion is something to be concerned about,” said Wong, who holds the Canada Research Chair in Medical Imaging Systems.
The AI software in the device was shown tens of thousands of images until it was able to recognize early-stage melanoma on its own. When caught early, melanoma is quickly treated. Without early detection, it is deadly.
The technology was developed by a research team that Wong oversees — the Vision and Image Processing Research Group. To commercialize it, Wong and two colleagues, Farnoud Kazemzadeh and Iman Khodadad,
started Elucid, which is part of Velocity, the university’s startup support program in the Tannery building in downtown Kitchener.
When it comes to developing new applications for artificial intelligence and machine learning, Wong and his team found everything they needed in Waterloo Region. The region’s AI cluster here seldom takes centre stage, but the technology is used by a growing number of startups.
“We have a huge amount of strength in AI in the region, particularly in the realm of operational AI, which is to take AI and make it work in operational scenarios,” said Wong. “Everything we need is here.”
Artificial intelligence will be central to the next round of big innovations, from self-driving cars to changing traffic lights according to traffic volume or hand-held scanners that detect early-stage skin cancer.
UW will not be left out of the AI boom, said Pearl Sullivan, the dean of engineering.
“AI at the University of Waterloo is both in the faculty of mathematics and the faculty of engineering,” said Sullivan. “We are very strong I think in both the fundamental research, as well as application of AI.”
Research groups at UW that are involved in AI include the Centre for Pattern Analysis and Machine Intelligence, the Artificial Intelligence Group and the Centre for Theoretical Neuroscience. Startups and tech firms in Waterloo Region that are using artificial intelligence and were founded by UW professors or graduates include Miovision, Vidyard, Sortable, Clearpath Robotics, D2L, Thalmic Labs, Kik Interactive and Hedgehog Medical. And there are more on the way.
“We have at least 15 startups in Velocity that are using AI,” said Sullivan, who adds that UW plans to expand its AI research.
In January, Maluuba, a Waterloo startup that developed an AI research lab, was acquired by Microsoft. Several months later, Maluuba closed its Waterloo office and moved its 50 employees to Montreal. Maluuba, which specializes in natural-language processing, was happy to make the move, said Kaheer Suleman, the company’s co-founder and chief technology officer.
Yoshua Bengio, head of the Montreal Institute for Learning Algorithms, became an adviser to Maluuba in January. Both the University of Montreal and McGill University are known for their research into AI.
“So it made sense to us at the time to go where the talent was,” said Suleman.
If Waterloo Region wants to be competitive in the field of artificial intelligence, it needs to put more money into it, he said.
“I do think that more emphasis by the university could be put on machine learning, and data science,” said Suleman.
“Especially at the graduate level.”