Matthew Tierney

�鶹��Ƶ engineering students encouraged to consider sustainability when designing future AI systems

Christopher.Sorensen — Mon, 16 Sep 2024 15:10:13 +0000

�鶹��Ƶ engineering students encouraged to consider sustainability when designing future AI systems

Christopher.Sorensen Mon, 09/16/2024 - 11:10

Cutline

Professors Hans-Arno Jacobsen, left, and Natalie Enright Jerger of the Faculty of Applied Science & Engineering are part of a team that is training computer system designers to integrate sustainability practices into the development of AI and machine learning systems (photo by Jenny Lee)

Matthew Tierney

Topic

Our Community

Computer & Electrical Engineering

Faculty of Applied Science & Engineering

NSERC

Sustainability

Subheadline

The Sustainable Data Systems for Data Science initiative aims to align Canada’s tech investments with its vision for a carbon-neutral future

The growing global adoption of artificial intelligence technologies, including machine learning, has created a new sustainability challenge: AI systems are energy-intensive – and the more sophisticated they become, the more resources they require.

To help address the issue, a team of experts including the University of Toronto’s Hans-Arno Jacobsen and Natalie Enright Jerger – both professors in the Edward S. Rogers Sr. department of electrical and computer engineering at the Faculty of Applied Science & Engineering – are launching an initiative that aims to align Canada’s tech investments with its vision for a sustainable, carbon-neutral future.

The Sustainable Data Systems for Data Science (SDSDS) project aims to train a new generation of computer and data scientists who can combine comprehensive technical skills with sustainability awareness. The project recently received a $1.6-million Collaborative Research and Training Experience (CREATE) grant from the Natural Science and Engineering Research Council of Canada (NSERC).

“A majority of students are unaware of how sustainability informs responsible development of platforms and systems because, frankly, there just aren’t many courses or learning paths available to them in this area,” says Enright Jerger, who holds the Canada Research Chair in Computer Architecture and is director of the division of engineering science.

“Another crucial component of this project is to equip our students with knowledge transfer strategies so they can seed these ideas in the workforce. Today’s trainees becoming tomorrow’s trainers.”

The SDSDS team propose training future computer systems designers to adopt a green approach when developing data analytics platforms and systems. The approach would apply to all aspects of the life cycle of development and deployment – such as hardware infrastructure, software systems and application domains.

To advance knowledge and dialogue on the issue, SDSDS will look to hold cross-university seminars on sustainability challenges and will offer courses on sustainable data science, along with summer school programs focused on energy-efficient software and hardware platforms.

The researchers also aim to connect students with industry via internships and applied research projects, enabling them to gain exposure to current challenges and facilitate cross-pollination of knowledge between industry and academia.

“The awareness about AI’s environmental impact is growing, but there is an expertise gap on how to address this very real problem – not just after the fact, but at inception. SDSDS aims to bridge this gap and prepare for the future resource demands of AI-driven industries,” says Bettina Kemme, professor of computer science at McGill University and team lead at SDSDS, which also includes Semih Salihoğlu of the University of Waterloo, Oana Balmau of McGill University and Essam Mansour of Concordia University.

“The potential of AI and machine learning systems are seemingly limitless,” says Jacobsen, who is the Jeffrey Skoll Chair in Computer Networks and Innovation. “Yet the true genius lies in building machine learning systems founded on sustainability principles. That’s real innovation.”

�鶹��Ƶ researcher leads effort to protect power utilities from quantum attacks

Christopher.Sorensen — Thu, 23 May 2024 20:12:49 +0000

�鶹��Ƶ researcher leads effort to protect power utilities from quantum attacks

Christopher.Sorensen Thu, 05/23/2024 - 16:12

Cutline

Faculty of Applied Science & Engineering researcher Deepa Kundur, second from right, is leading a collaboration between academia and industry that’s focused on developing solutions to protect power utilities from cyberattacks using quantum technologies (photo by Neil Ta)

Matthew Tierney

Topic

Our Community

Electrical & Computer Engineering

Faculty of Applied Science & Engineering

Quantum

Quantum Computing

Research & Innovation

Sustainability

Subheadline

“Technology is always changing the threat landscape. And quantum computing, which is becoming more feasible and practical, is a powerful tool that will make our classical defences obsolete”

A researcher from the University of Toronto is leading a multidisciplinary research group that aims develop quantum-based technology solutions to defend power utilities against future cyberattacks.

With the support of a first-of-its-kind NSERC Alliance-Mitacs Accelerate grant worth $1.45 million, the group is working at the intersection of quantum, cybersecurity and critical infrastructure.

“We have to stay ahead of the game,” says group lead Deepa Kundur, professor and chair of �鶹��Ƶ’s Edward S. Rogers Sr. department of electrical and computer engineering in the Faculty of Applied Science & Engineering.

“Technology is always changing the threat landscape. And quantum computing, which is becoming more feasible and practical, is a powerful tool that will make our classical defences obsolete.”

Kundur’s project is a collaboration between academia, Hydro-Québec and Xanadu, one of Canada’s most successful quantum computer startups. A second team – headed by Associate Professor Atefeh Mashatan of Toronto Metropolitan University and involving quantum solution leaders Crypto4A and evolutionQ – will build a road map for the classical-to-quantum migration for power grids in preparation for a future transition.

Quantum enhancement is the next stage in the evolution of today’s smart grids, so-named because they incorporate information-communication technology (ICT) into their operations. ICT has allowed smart grids to adapt to changing conditions and electricity load, as well respond more efficiently to natural disasters in order to meet society’s increasing power needs in an intelligent, sustainable way.

“ICT and its advanced sensors generate more data than before,” says Kundur. “We transport this data to different parts of the grid to start co-ordinating information to make decisions based on synchronized information and enhanced situational awareness.”

One potential downside of a data-driven smart grid, however, is the introduction of new vulnerabilities since attackers can now target not just the physical infrastructure, but the information that flows through it.

That’s because a smart grid’s connectivity increases opportunities for access. Also, ICT adds a level of complexity that results in emergent properties that are difficult to predict and can be challenging to safeguard. And the standards and policies put in place to mitigate operational variations mean there’s a level of interoperability between working grids that hackers can use to their advantage.

While cybersecurity experts have so far incorporated layers of defences into our smart grids, Kundur warns that those safeguards are not ready for quantum technologies.

“Algorithms and cryptography that are incredibly difficult for classical computers to crack become solvable with a quantum computer,” she says. “And then other questions arise. For example, when the power utilities themselves start to use quantum sensors, is this quantum-enhanced information better for attack detection or does it give attackers an ability to hide themselves?”

The question is tough to answer when you consider that quantum sensors of this nature – and the quantum data they would generate – don’t exist yet.

“We’ll take classical data, use models to predict what quantum versions of the information would appear to be, and then perform anomaly and attack detection on it,” says Kundur.

“We’ll be experimenting with quantum machine learning for better pattern recognition to detect a cyberattack. This is a highly exploratory project.”

Even if it’s decades before manufacturers integrate quantum attack-detection algorithms in their devices, Kundur says foundational research that she and her team will carry out in the next few years is a valuable endeavour.

“Security is a process. It’s very much a dynamic interaction,” she says. “And though we can never get to 100-per-cent protection, it’s something we have to continually try to achieve.”

�鶹��Ƶ researcher's AI model could help optimize e-commerce sites for users who are colour blind

Christopher.Sorensen — Mon, 15 Apr 2024 17:23:10 +0000

�鶹��Ƶ researcher's AI model could help optimize e-commerce sites for users who are colour blind

Christopher.Sorensen Mon, 04/15/2024 - 13:23

Cutline

Parham Aarabi, an an associate professor of electrical and computer engineering, built an AI model that simulates how users interact with images on an e-commerce site – including those experiencing colour blindness (photo by Matthew Tierney)

Matthew Tierney

Accessibility

Artificial Intelligence

Electrical & Computer Engineering

Faculty of Applied Science & Engineering

Research & Innovation

Subheadline

Parham Aarabi found that, in general, users experiencing colour blindness are 30 per cent more likely to click on a monochrome image

University of Toronto researcher Parham Aarabi has created an artificial intelligence model that mimics how users navigate e-commerce websites – and it may be able to help retailers optimize their sites for people experiencing colour blindness and other conditions.

Called PRE, the AI-generated tool sees virtual users browse, pause on a page, add items to cart and click on discounted items.

While the tool shows that users tend to be drawn to colourful images, Aarabi also wanted to see how those experiencing full and partial colour blindness might respond.

“Around eight to 10 per cent of the population has a type of colour-blindness,” says Aarabi, an associate professor in the Edward S. Rogers Sr. department of electrical and computer engineering in the Faculty of Applied Science & Engineering. “There are a number of ways the eye can be confused by colour, commonly between red and green or blue and yellow.

“I wanted to see how this might impact web navigation.”

Aarabi set up an experiment. He altered a retail clothing website to simulate how it would appear to someone with protanomaly, or a reduced ability to perceive red light. One might think of it as applying a filter, or lens, which Aarabi then modified to approximate eight other variations of colour deficiency.

For each variation, Aarabi initiated one million navigation sessions with AI virtual users and tracked the image click rates. He found that, in general, someone with colour-blindness is 30 per cent more likely than a colour-abled user to click on a monochrome image. The results will be presented in a paper at the 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society this summer.

A screenshot shows all nine versions of Aarabi’s test website, each filtered to simulate a variation of colour-blindness. The bottom-right version is weak protanomaly, or a reduced ability to perceive red light (image by Parham Aarabi)

The boost factor that website designers count on with colour doesn’t translate to everyone, notes Aarabi.

“When people are designing sites or presenting products, they need to stay cognizant that eight per cent of the population is not going to be impacted. You need to add better descriptions and more textual information to guide users through the shopping process.”

Aarabi sees this study as one of many that can benefit from PRE, whose neural net took two years to train with data from 110,000 real-life user sessions.

“To measure its accuracy, we set up a sample site and predicted what actions the AI virtual users will take – what percentage would add to cart, what percentage would buy a particular product, and so on – and also ran a test of the site with people,” says Aarabi. “PRE correctly mimics a human user’s actions 90 per cent of the time.”

There are benefits with using AI virtual users for a study. One can run experiments more quickly, on a larger scale, and can recreate as many sessions as desired. The AI model eliminates the need, for example, to locate and coordinate many thousands of willing colour-blind participants.

Aarabi has plans to use PRE to test other barriers to accessibility, such as dyslexia or motor impairments. His long-term goal is provide an auditing service for companies that allows them to test a web design’s impact on users with various conditions before or after launch.

Such goals are part of Aarabi’s research effort to mitigate negativity about AI.

“There’s a lot of worry, even within the tech community, about AI taking over or replacing us in some capacity,” he says. “If we can make AI more humanlike in some way, build in some empathy and have it mirror the reactions that humans have, we could dispel some of those concerns.”

“Professor Aarabi has been a pioneer in the application of AI, from past research cautioning against bias in training data sets to this current project, which uses the AI advantage to address accessibility issues,” says Professor Deepa Kundur, chair of the department of electrical and computer engineering. “Parham brings a valuable, forward-thinking approach to leveraging AI for positive outcomes.”

�鶹��Ƶ researcher’s data-driven platform aims to predict when emergencies will happen

Christopher.Sorensen — Fri, 22 Mar 2024 15:02:23 +0000

�鶹��Ƶ researcher’s data-driven platform aims to predict when emergencies will happen

Christopher.Sorensen Fri, 03/22/2024 - 11:02

Cutline

(photo by Matthew Tierney)

Matthew Tierney

Topic

Our Community

Cities

Electrical & Computer Engineering

Faculty of Applied Science & Engineering

Research & Innovation

Subheadline

Alberto Leon-Garcia is collaborating with Edmonton Fire Rescue Services and TELUS to support first responders in Alberta's second largest city

A University of Toronto researcher is working with Edmonton Fire Rescue Services and TELUS, through its Data for Good program, to predict when emergencies are likely to occur in Alberta’s second largest city.

The tool being developed by Alberto Leon-Garcia, a professor in the Edward S. Rogers Sr. department of electrical and computer engineering in the Faculty of Applied Science & Engineering, and the two partners leverages data to more efficiently allocate municipal emergency resources and help first-responders.

Leon-Garcia says many emergency events can be predicted because people’s behaviours tend to follow certain patterns.

“The pulse of the city is driven by people and their activity,” he says, “and their activity exhibits seasonality.”

Leon-Garcia’s platform uses data from 11 years of emergency calls, which provide the time and approximate location of each event as well as the type of emergency – house fire, medical emergency, traffic accident and so forth – in addition to other relevant data points.

“For the city of Edmonton, we look at the neighbourhood level, at demographics, land use, transportation capabilities, population density,” says Leon-Garcia. “We consider the timing of the events, how they vary by season, month, day of the week, hour.

“This can allow you to predict the rate of events in the vicinity of each fire station in the next week or month, for example. Right there, that’s a valuable input to resource allocation – how many trucks, how many people you assign and where.”

Creating the model required collecting the necessary data and then refining it so it was free from errors and standardized, possibly transformed or aggregated. Next, researchers needed to determine the most useful way to analyze it.

“Deep neural networks were not appropriate in this instance,” says Leon-Garcia, referring to the machine learning techniques behind such tools as ChatGPT. “You can try – and we did – but we did not have the volume of data to train a neural network.”

Instead, he turned to “well-established advanced analytics.”

The data analysis will generate various graphs, heat maps and other tables that display the type and mixture of emergency events that the model considers normal in and around Edmonton for a given time and place while taking into account variables such as weather.

By following events in real time and comparing them to what is anticipated, researchers can detect anomalies and potential vulnerabilities in the model.

“For example, one time we noticed that the fire event numbers in a neighbourhood didn’t correspond to the models,” says Leon-Garcia.

“It was later confirmed that an arsonist was active during that period.”

Over the years, Leon-Garcia has applied his predictive models to various road transportation systems, including in Toronto and the San Francisco Bay Area. He has also applied his anomaly detection systems to detect faults in computer networks and cyberattacks.

Given that each partner in such a project typically has its own goals and unique data collection processes, Leon-Garcia says it’s critical to take a collaborative approach.

“You can’t come in and say, ‘I have this neat platform, you have to change the way you do things,’” he says. “It doesn’t work that way. You have to pull together, factor in their long-term goals, their privacy concerns, their flexibility. They generally see the usefulness of the approach and [then] it’s more a question of how you get from here to there.”

Professor Deepa Kundur, chair of the electrical and computer engineering department, says Leon-Garcia has consistently demonstrated how data streams hold the key to creating smarter, safer cities.

“His partnership with Edmonton FRS and TELUS has the potential to greatly enhance life-saving initiatives and will, no doubt, serve as a catalyst for future collaborations.”

�鶹��Ƶ researchers develop rapid MRI technique for better cancer detection and therapy

Christopher.Sorensen — Fri, 23 Feb 2024 19:30:43 +0000

�鶹��Ƶ researchers develop rapid MRI technique for better cancer detection and therapy

Christopher.Sorensen Fri, 02/23/2024 - 14:30

Cutline

(photo by Willie B. Thomas/Getty Images)

Matthew Tierney

Topic

Breaking Research

Institute of Biomedical Engineering

Cancer

Electrical & Computer Engineering

Faculty of Applied Science & Engineering

Research & Innovation

Subheadline

“We create images every second, and sometimes less, and those images are not going to suffer from low resolution”

Researchers at the University of Toronto’s Faculty of Applied Science & Engineering have developed a rapid magnetic resonance imaging (MRI) technique to help doctors better detect and diagnose tumours.

The new approach – by Hai-Ling Cheng, a professor in the Institute of Biomedical Engineering and the Edward S. Rogers Sr. department of electrical and computer engineering, and PhD candidate Alex Mertens – could provide physicians with guidance during surgery and other therapeutic interventions.

Based on novel analysis of raw patient data collected from imaging sessions with standard MRI equipment, the algorithm Cheng and Mertens developed reduces the duration between each image acquisition from more than 20 seconds to one second without sacrificing image sharpness.

“People in the field have been trying to get high spatial resolution concurrently with temporal resolution for the past 25 years,” says Cheng.

Cheng and Mertens, with the help of the �鶹��Ƶ Innovations and Partnerships Office, have applied for a patent and are partnering with companies to bring their MRI technique to market.

“In practice, doctors always follow up imaging results with a biopsy for definitive confirmation to more accurately determine the grade of cancer and its stage,” Cheng says.

“Our technique is not meant to displace the biopsy. But by better characterizing the underlying pathology at the vascular and cellular level, we can mitigate randomness in the sampling when the doctor goes in with a biopsy needle.”

Professor Hai-Ling Cheng, pictured, and ECE PhD candidate Alex Mertens have developed a novel method to analyze data acquired from magnetic resonance imaging (photo by Matthew Tierney)

MRI is used to scan soft tissues like muscle or fat because it offers the best contrast compared to other modalities such as X-rays and ultrasounds. The contrast allows doctors to discern different cell types and identify small cancerous growths.

“Let’s say you’ve got a liver and a kidney, and you want to image them both in the same area of the body,” says Cheng. “If you were to take an X-ray, you would get one contrast level – one grey scale specific to the liver and one grey scale specific to the kidney.”

With MRI technology, however, there’s different physics at play that allows for refined gradients. An MRI scanner produces a strongly magnetized field inside the patient’s body into which it pulses a radio frequency, or RF, wave. The wave affects the water protons in soft tissues, which react to the pulse and emit signature return signals.

“The data from the return signal doesn’t tell you the shape of an object but the frequency content of the object,” says Cheng. “We structure that return RF signal into a matrix, which we can then convert into a high resolution image.”

Researchers can change the magnetic strength and the frequency of the pulse to obtain different contrasts, much like a music producer can increase and decrease the volume of individual tracks in a song.

To enhance the RF signal further, a contrast agent is intravenously injected into the patient beforehand: usually gadolinium, which is non-radioactive. The dynamics of the gadolinium distribution – that is, the speed of its uptake and washout in cells – give doctors additional information about the malignancy of the tumour.

“Tumours not only have a larger blood volume, but because their blood vessels are very messed up and tortuous, they also tend to be very, very leaky,” says Cheng.

However, the MRI procedure is notoriously slow. The scanner must repeatedly acquire frequency domain data at different coarse and fine-grain resolutions. Typical temporal resolution is 20 seconds and gadolinium washout in a tumour can take as little as 10 seconds.

“Typically, it takes 256 acquisition lines to create one image,” says Cheng. “Rather than reconstructing a full image every 20 seconds or every minute – that’s kind of pointless, because you’re missing the dynamics – our algorithm extrapolates information based on successive sampling of just one acquisition line.

“We create images every second, and sometimes less, and those images are not going to suffer from low resolution.”

“The work that Hai-Ling and her team are doing is a testament to how electrical and computer engineering technologies can impact the health-care sector,” says Professor Deepa Kundur, chair of the electrical and computer engineering department in the Faculty of Applied Science & Engineering. “Hai-Ling and Alex have spent years building on their knowledge of MRI physics and human biology, demonstrating how interdisciplinary perspectives in engineering save lives. Well done.”

EV systems course prepares �鶹��Ƶ students for fast-growing field

Christopher.Sorensen — Thu, 25 Jan 2024 19:29:15 +0000

EV systems course prepares �鶹��Ƶ students for fast-growing field

Christopher.Sorensen Thu, 01/25/2024 - 14:29

Cutline

Sessional lecturer Zhe Gong looks over the components that make up the electric vehicle lab station in the Faculty of Applied Science & Engineering’s Energy Systems Lab (photo by Matthew Tierney)

Matthew Tierney

Topic

Our Community

Electric Cars

Electrical & Computer Engineering

Faculty of Applied Science & Engineering

Sustainability

Subheadline

“Our objective is to prepare our students to be innovation leaders to meet society’s needs”

The University of Toronto is offering a graduate course in electric vehicle systems that combines a theoretical background in power and energy flow with hands-on experience.

As demand grows for automotive engineers in the fast-growing electrification field, the multidisciplinary course –offered through the Edward S. Rogers Sr. department of electrical and computer engineering in the Faculty of Applied Science & Engineering – aims to give graduate students a solid understanding of the concepts needed to design high-performance EV systems. It also discusses EV subsystems, with a focus on energy modelling and efficiency.

The new course, launched in fall 2023, comes amid efforts to build Stellantis-LGES and Volkswagen battery plants in Ontario, creating the need for more engineers with EV technology skills.

“The grad course was a pilot to explore how a hands-on EV course could fit into our curriculum,” says alumnus Zhe Gong, the course’s sessional lecturer.

“We’re making sure that we understand the hardware requirements, that we can frame a meaningful scope of experiments and that students can experience sufficient hands-on time through the lab sessions.”

The number of EV models available on the market doubled from 2018 to 2022 to a total of 500, according to the International Energy Agency. And the number of zero-emission vehicles (ZEVs) on the road is slated to increase dramatically in the next decade. Within Canada, this will be spurred in part by the federal government’s emissions reduction plan, which would require ZEVs to comprise all light-duty vehicles sales by 2035.

“Our objective is to prepare our students to be innovation leaders to meet society’s needs,” says Professor Deepa Kundur, chair of the electrical and computer engineering department. “This is especially the case when it comes to sustainability, and a new course offering in electric vehicles helps build a robust talent pipeline to provide the electrification industry with people ready to make a difference.”

Gong says that the lab setup, situated in the department’s undergraduate Energy Systems Lab, took six months to develop and incorporated research from the University of Toronto Electric Vehicle (UTEV) Research Centre into the hardware and software requirements. The setup includes a dynamometer that simulates how the road applies loading force to the vehicle propulsion system, a lithium-ion battery, hardware switches to selectively connect the battery to motor and charger – and a power supply to act as the on-board charger.

EV working components are arranged on lab tables to provide full access, “as if a bench-top electric vehicle,” says Gong.

“We also have something that’s quite unique for an EV lab – EV supply equipment, something you would actually see in a home. We customized the connection between our charger and power distribution panel to allow students to step through the communications interface required for the vehicle to engage the charging sequence.”

Components in the EV experimental lab setup (photo by Zhe Gong)

“Over the years, a lot of the equipment in the Energy Systems Lab has been designed by profs, research associates, grads, as well as undergrad students,” says Afshin Poraria, director of teaching labs, in reference to the collaborative approach between faculty members, UTEV and undergrad lab managers in creating the lab setup.

“We build whatever we can to feed and expand the student experience. Before long, we’ll likely have undergrad students using this equipment.”

UTEV Director Olivier Trescases, a professor of electrical and computer engineering, says that the “all-hands-on-deck” approach and the creative solutions to equipping the labs are necessary to provide the best possible learning environment for students.

“Over the past few decades, ECE has made major investments to design and deploy custom infrastructure to deliver a unique training experience that is simply not possible with off-the-shelf equipment.”

With three degrees from �鶹��Ƶ Engineering, Gong says he’s happy to play a role in the course.

“Through my undergrad days to now, I’ve always known ECE’s labs to be collaborative ones. It’s very exciting for me to be contributing to a new stage, building custom equipment and now teaching the students about electric vehicles – something that I’m passionate about,“ he says. “It’s like coming full circle.”

�鶹��Ƶ course on brain-machine interfaces introduces undergrads to next-gen health care

Christopher.Sorensen — Fri, 05 Jan 2024 15:18:55 +0000

�鶹��Ƶ course on brain-machine interfaces introduces undergrads to next-gen health care

Christopher.Sorensen Fri, 01/05/2024 - 10:18

Cutline

Assistant Professor Xilin Liu, standing, guides students Mona Murphy, left, and Nishant Kumar, right, as they analyze Kumar’s EEG (photo by Matthew Tierney)

Matthew Tierney

Topic

Our Community

Institute of Biomedical Engineering

Temerty Faculty of Medicine

Electrical & Computer Engineering

Faculty of Applied Science & Engineering

Undergraduate Students

University Health Network

Subheadline

“You start to look at the brain differently, the mechanisms of its disorders, what clinicians and neuroengineers are doing to treat them”

Undergraduate students taking a new University of Toronto course have to use their brains in more ways than one.

Called Interfacing and Modulating the Nervous System (ECE441), the course in the Edward S. Rogers Sr. department of electrical engineering and computer engineering in the Faculty of Applied Science & Engineering introduces fourth-year students to neuromodulation, a multidisciplinary area that draws on knowledge about signal processing, control theory, electronics and machine learning.

Devices employing neuromodulation deliver therapeutic stimulation to targeted areas of the brain and are used to treat a range of conditions, including chronic pain, neurological disorders such as Parkinson’s disease and epilepsy, depression, spinal cord injuries and hearing or vision loss.

During one recent lab session, students Jannis Gabler and Aurora Nowicki said they were amazed by what they could learn after they hooked up a team member.

“We were measuring electrical activity in his visual cortex,” Gabler says. “Just by extracting data from his signature, we could tell whether his eyes were closed or not.”

The course was developed through the combined efforts of faculty members Xilin Liu and Ervin Sejdić, as well as the CRANIA Neuromodulation Institute. Support from the department included the purchase of specialized equipment – biosensing headsets that allow students to acquire an electroencephalograph (EEG) – and technical support from Afshin Poraria, director of teaching labs, and lab manager Iman Makhmal Koohi.

Liu, an assistant professor in the electrical and computer engineering department, says hands-on work is a crucial component of the course, which exposes students to neural interfacing techniques and applications at a time when many are considering graduate research or starting their career.

While he notes that universities such as MIT and Stanford University offer similar courses, he says they tend to be geared toward graduate students.

“Introducing such a course at the ECE undergraduate level is quite a unique approach,” he says. “By leveraging the strengths of ECE labs, we implemented hands-on experiments enabling students to collect, analyze and modulate their own EEG signals in real-time.”

A student holds a biosensing headset during a lab session (photo by Matthew Tierney)

The course material combines Liu’s teachings on electronics and neural interfacing technology with Sejdić’s work on signal processing, control theory and machine learning. It also includes guest lecturers with various clinical expertise. They include Milos Popovic, a professor in the Institute of Biomedical Engineering (BME), and Taufik Valiante, a scientist at the University Health Network and an associate professor of surgery in the department of surgery in the Temerty Faculty of Medicine with a cross appointment at BME – both early supporters of the course.

During off-campus excursions to nearby hospitals and clinical settings, such as Toronto Western Hospital, the students hear from neurosurgeons and neuroscientists about real-world advancements in this field.

“The first lecture was one of the most interesting I’ve ever had at �鶹��Ƶ,” Nowicki says.

“We learned early on how little we understand the brain. It’s incredibly complex,” adds Gabler.

Liu says we may never have an accurate general model of the brain, which means therapeutic determinations must be considered on patient-by-patient basis – a challenge for clinicians.

“You cannot keep people in the hospital for a long time just to monitor the progression of the disease,” Liu says. “And optimization done at the hospital might only be calibrated for the specific time of day, or might not work when they go back home, or while asleep.”

But wearable or implantable neural devices that use machine learning are ideally suited to address this challenge because they collect large amounts of data from patients over long periods of time. With this data, machine learning can learn the optimal timing and parameter configuration.

The field is growing so fast that the electrical and computer engineering department looks forward to expanding its course offerings in neurotechnology, says Deepa Kundur, a professor and department chair, who adds that this course helps build a strong foundation in the field.

“It’s an excellent example of how bringing awareness of cutting-edge applications into the classroom setting, through access to research labs and a diverse instructor team, allows students to see the incredible potential opportunities for their electrical and computer engineering skill set,” she says.

The department also plans to make equipment available to students outside the class. The instructors are working with the student club NeuroTECH to organize workshops and hands-on sessions.

Nowicki, for one, recommends the course to fellow students even if they’re not considering a future in health care.

“You start to look at the brain differently, the mechanisms of its disorders, what clinicians and neuroengineers are doing to treat them,” she says. “And so many people have friends or a family member who has experienced a disorder.”

�鶹��Ƶ Engineering professor aims to reimagine the internet

Christopher.Sorensen — Thu, 16 Nov 2023 20:52:57 +0000

�鶹��Ƶ Engineering professor aims to reimagine the internet

Christopher.Sorensen Thu, 11/16/2023 - 15:52

Cutline

Professor J.J. Garcia-Luna-Aceves aims to forge a smarter, more equitable internet (photo by Matthew Tierney)

Matthew Tierney

Topic

Our Community

Canada Excellence Research Chairs

Electrical & Computer Engineering

Faculty of Applied Science & Engineering

Research & Innovation

Subheadline

J.J. Garcia-Luna-Aceves, who has been awarded a Canada Excellence Research Chair, says the internet runs on 50-year-old technology that hasn’t kept pace with advancements in other fields

For J.J. Garcia-Luna-Aceves, the networks that make up the internet – both the physical layer of routers and switches, as well as the protocols and algorithms that distribute data – hold unused intelligence with the potential to foster major advances.

“The internet we are using today was designed 54 years ago. I mean, I wouldn’t drive a 50-year-old car. But that’s what people are doing,” says Garcia-Luna-Aceves, a professor in the University of Toronto's Edward S. Rogers Sr. department of electrical and computer engineering in the Faculty of Applied Science & Engineering. “There have been adjacent technological advancements over the years in radios, machine learning and the like. But the main algorithms used in the internet protocol stack haven’t followed suit.”

Garcia-Luna-Aceves was this week appointed a Canada Excellence Research Chair in Intelligent Digital Infrastructures, a position that will help him explore and tap into the resources that lie “under the hood” of computer networks. Awarded by the Government of Canada, the Canada Excellence Research Chair (CERC) program aims to fund ambitious research projects that help position Canada as a global leader in innovation.

Garcia-Luna-Aceves points to the smartphone’s upward trajectory as a counterexample to that of the internet.

“Your smartphone is intelligent,” he says. “What makes it so? To act intelligently, one needs memory. Endowing network entities with vast amounts of memory capacity would allow us to rethink the routing protocols that are the backbone of the internet.”

Garcia-Luna-Aceves himself is responsible for sections of that backbone. In the late 1980s, he developed an algorithm to handle the fast communication demands of a flying network of missile interceptors. Although the algorithm was never used for that purpose, it was adopted by networking and communications giant Cisco as part of their Enhanced Interior Gateway Routing Protocol (EIGRP), called DUAL, which is still widely used today.

“That technology was great for that time,” says Garcia-Luna-Aceves. “But we need to build routing tables that not only take into account policies, but the characteristics of the use – the content and the service being accessed and the intent for the communication.”

To get there, he says, we need to go back to first principles – which means forgetting the hardware and bandwidth limitations of 50 years ago, and asking ourselves how we would design the internet today.

That means we need to look at whom the technology is enabling and whom it is not.

“Any solution needs to be not only scalable but affordable and deployable in different settings,” he says. “There are areas, communities and industries today that the internet does not reach, where it has potential as a technology enabler – applications in agriculture, in fishery, you name it. And we can only learn about them by talking to those who have been disenfranchised in the past.”

Garcia-Luna-Aceves plans to launch a multidisciplinary research hub, called the Centre of Excellence for Networking Innovation in Toronto, that draws on international researchers and industry partners to further his vision.

“Professor Garcia-Luna-Aceves has his sights on an ambitious and impactful project. Everything in his distinguished career suggests he’s the person for the job,” says Professor Deepa Kundur, chair of the department of electrical and computer engineering. “I enthusiastically congratulate him on his CERC, a well-deserved recognition of his incredible talent.”

“It’s an opportunity like very few others,” says Garcia-Luna-Aceves. “It’s a clean slate to pursue my research. That I can do so in one of the best, most inclusive cities in North America is an added blessing.”

�鶹��Ƶ researchers build online glaucoma simulator

Christopher.Sorensen — Thu, 21 Sep 2023 14:26:48 +0000

�鶹��Ƶ researchers build online glaucoma simulator

Christopher.Sorensen Thu, 09/21/2023 - 10:26

Cutline

An online tool built by Professor Willy Wong and PhD candidate Yan Li of the Faculty of Applied Science & Engineering depicts glaucoma from a patient’s perspective, with a baseline image on the left and a moderate defect at right.

Matthew Tierney

Topic

Breaking Research

Faculty of Applied Science & Engineering

Subheadline

“If you do an internet search for what glaucoma looks like, the images ... are tunnel vision with the periphery blacked out. There’s very little truth to this."

Researchers at the University of Toronto have built an online simulator that gives an improved visual representation of the deterioration caused by glaucoma.

The simulator, built by Professor Willy Wong and PhD candidate Yan Li of the Faculty of Applied Science & Engineering and their collaborators, depicts the disease from the patient’s perspective.

Almost all other representations on the internet are inaccurate, says Wong.

“If you do an internet search for what glaucoma looks like, the images returned are tunnel vision with the periphery blacked out. There’s very little truth to this.” says Wong of the Edward S. Rogers Sr. department of electrical and computer engineering. “What’s really happening is patches of your visual field are losing their spatial integrity – more what you might see when you just wake up, when you’re not really focused in.”

Wong and Li’s online simulator is based on a data-driven model they developed to help quantify glaucoma measurements. Their model, which takes into consideration the physiological mechanism of the eye, was recently published in a paper in the journal Translational Vision Science & Technology.

“We worked closely with glaucoma specialists – two are co-authors on the paper – and they helped us get up to speed with the pathophysiology of the eye,” Li says. “Working side by side with them, we got precious first-hand experience about what clinicians need from technology.”

Glaucoma tends to affect older individuals and is generally painless. The disease is characterized by elevated pressure in the fluid that fills the eyeball, which then compresses and damages the nerve endings that transmit signals to the brain.

PhD candidate Yan Li says he and Professor Willy Wong worked closely with glaucoma specialists (photo by Matthew Tierney)

The disease is a leading cause of blindness, but its progression can be greatly slowed with medication and other interventions. To determine when these are needed, doctors routinely perform proactive monitoring, which includes several qualitative tests, such as examining the optic nerve or the retinal layer. These typically employ numbered measurements to be as precise as possible.

“But even with these measurements, the doctors don’t always know if it’s trending upwards or downwards or staying the same,” says Wong.

This is because many of the physiological measurements are inherently noisy. For example, visual field tests rely on the measurement of visual thresholds – that is, the minimal amount of energy required to see – but these thresholds can be unreliable. The noisiness is also compounded by the many years over which the tests are taken and the gaps where appointments are inevitably missed.

Li says that what makes this model unique is the way it combines data from these tests with knowledge about the biology of the eye.

“You have to be mindful of how the nerves go from the eye itself – from the visual field into the optic disc,” he says. “If you know the relationship between the two and add that causality to the clinical data, you have a much better prediction tool.”

Wong and Li’s online simulator allows the user to set the patient’s age range and control progression rates from a starting point of mild, moderate or severe. It then simulates the yearly progression of the disease through photos of people, landscapes and city scenes.

“I showed an ophthalmologist our simulator and he said, ‘This is exactly what I need,’” says Wong. “Glaucoma is so slow in developing, it’s apparently hard to convince patients to take the medication because they don’t see or understand the difference.”

Li’s next step in this project is to combine different testing methodologies for glaucoma to shorten the time required to reliably detect its onset using machine learning (ML) methods. Faster detection will allow for sooner intervention and a better chance at stopping the irreversible loss of vision, he says.

“The development of medicine relies heavily on doctors passing their experience to one another, and ML helps to quickly and efficiently learn and transfer the knowledge from domain experts,” he says. “This is a major benefit of the era of big data in health care.”

“Combining computing power with deep knowledge of biology, such as Wong and Li have done, is very compelling,” says Professor Deepa Kundur, who chair of the electrical and computer engineering department. “The results speak for themselves, and I wouldn’t be surprised to see this hybrid model make a difference in other applications.”

�鶹��Ƶ graduate students use AI to improve imaging tool used during breast cancer surgery

Christopher.Sorensen — Fri, 11 Aug 2023 20:15:24 +0000

�鶹��Ƶ graduate students use AI to improve imaging tool used during breast cancer surgery

Christopher.Sorensen Fri, 08/11/2023 - 16:15

Cutline

Bryant Bak-Yin Lim, left, and Ali Yassine simulate reviewing a breast cancer tissue scan. As MITACS interns at Perimeter Medical Imaging, Lim and Yassine developed new AI algorithms for breast cancer imaging (photo by Neil Ta)

Matthew Tierney

Topic

Our Community

Temerty Faculty of Medicine

Artificial Intelligence

Faculty of Applied Science & Engineering

Graduate Students

Research & Innovation

Subheadline

As part of an internship with a medical imaging firm, Bryant Bak-Yin Lim and Ali Yassine developed algorithms for a next-gen imaging system

Bryant Bak-Yin Lim and Ali Yassine got the chance to make a difference in the lives of patients last summer by improving how breast cancer surgery is performed.

The two researchers at the University of Toronto were participating in internships with Perimeter Medical Imaging, a company with offices in Toronto and Dallas, that were organized through Mitacs. There, they developed artificial intelligence (AI) algorithms for the next generation of a medical imaging system that helps surgeons visualize tissue microstructures during a lumpectomy to determine whether they have excised all the cancerous tissue.

Their algorithms prioritize suspicious images, making it easier for surgeons to parse the images and reduce time spent in the operating room.

The imaging device is about the size and shape of a small photocopier, says Lim, an MD student in the Temerty Faculty of Medicine who is completing an MD-oriented version of the master of engineering program in the Faculty of Applied Science & Engineering.

Situated within the operating area, the device employs a technology called optical coherence tomography (OCT), which is similar to ultrasound technology but uses light instead of sound to generate images, resulting in an image resolution 10 times greater than ultrasound.

OCT has been widely used in clinical settings, including ophthalmology, dermatology and interventional cardiology, but Perimeter’s device is the first to bring wide-field OCT imaging into the OR, Lim says.

“The tissue removed from the patient is put in a plastic bag and placed on a glass imaging plate on the device, using mild suction to hold it in place,” Lim says. “Light shoots up from the optical imaging system below, penetrates the tissue and reflects back into the device, which then displays results as a digital image on the monitor.”

Surgeons are looking for any suspicious features in what’s called the “margin,” striving for about a two-millimetre rim of healthy tissue along the outer edges of the excised tissue.

“Currently, to assess a margin, specimens are sent out to a pathologist. That process usually takes days,” says Yassine, who recently graduated with a master’s degree in electrical and computer engineering. “If there’s cancerous tissue left, patients sometimes have to go back for another procedure, with all the risks and resource costs that come with it.

“The type of deep learning algorithm that I trained, called a convolutional neural network, can analyze the tissue image and identify whether the material is suspicious or non-suspicious with a very high accuracy rate.”

The challenge then is to display this analysis for the surgeon so they can make a timely, informed decision on whether they need to return to the operating table and remove more tissue from the patient, who is still under anesthesia.

Lim was tasked with building an efficient user interface to guide the surgeon.

“This device typically outputs hundreds of images, and it’s challenging for a surgeon in the OR to read through all of them and make a decision on the spot,” he says.

“I developed an algorithm that clustered images together based on certain parameters and then displayed only the most representative one.”

The algorithm reduced the hundreds of images to a more manageable number of thumbnails that account for all the information gathered from the tissue scan. The surgeon can also manipulate the digital images to see the tissue from different perspectives.

There is great potential for AI-enhanced tools to make the medical professional’s work – and patient’s experience – smoother, says Ervin Sejdić, a professor in the Edward S. Rogers Sr. department of electrical and computer engineering who supervised both students.

“The Perimeter device that Bryant and Ali worked on is part of a wave of new tools that do the grunt work of sorting through and repackaging the copious amounts of data necessary for complex procedures or diagnoses,” says Sejdić.

“This helps doctors sharpen their focus on the treatment.”

For his part, Yassine didn’t expect he would be this interested in medicine before he undertook this internship. He is finishing up his master’s project – a multi-class labeller algorithm for Perimeter that identifies specific tissues in breast cancer samples – and is planning to continue his career in medical technology.

“I had my own personal health challenges a while back, and that has motivated me to work in this field,” he says. “It’s nice to help people through technology.”

Lim, who has two years left to complete his medical degree, says, “I hope to combine parts of AI and medicine and apply that to my future practice, whether industry research or some other collaborations. That’s where I want to bring my career to.”

“We are growing our MEng program in part because there are so many exciting possibilities out there for graduates,” says Professor Deepa Kundur, chair of the department of electrical and computer engineering.

“Lim and Yassine’s internships at Perimeter demonstrate how quickly hands-on training can translate into real-world results.”

Note: Technologies referenced in this article are currently not available for sale in the United States and have not been evaluated by the FDA.