A Helping Hand in Deciphering Health and History

Machine learning tools can sift through huge databases of photos and video, helping archive historical objects and interpret medical images.

 |  Transcript [PDF]

Cameras are everywhere these days: capturing everyday moments, recording exceptional events and objects, and helping doctors take a closer look at our bodies when we’re sick. But while we’re very good at collecting this information, it can be hard find a particular file later or to spot an abnormality in a medical scan.

Matthew Kyan, professor of electrical engineering and computer science at York University’s Vision: Science to Applications (VISTA) program, uses machine learning and pattern recognition to make photos and videos easier to retrieve when we need them.

Beyond home videos and selfies, Kyan’s interdisciplinary collaborations are making big contributions from historical archives and archeology to health and medicine.

“We work with filmmakers, working on cultural heritage types of projects,” says Kyan. “We’re looking through large archives of historical data. We also work with archeologists, digitizing and storing large collections of artifacts to build tools that can more effectively allow them to query these large databases.”

Digitizing a searchable database of important historical and archeological objects makes it possible for people all over the world to study them. 3D models even make it possible to interact with objects on a screen or using VR.

Kyan’s tools can also be used to analyze medical images for abnormalities, even when a medical professional isn’t sure what to look for.

“We work with doctors, looking for a tumour within a CT scan or being able to identify a valve within the heart that has been malformed,” explains Kyan. “Or, if it’s an MRI scan, to search for something within the brain that is pathological; it’s not something that would commonly be found, there’s not a model for how to find it.”

What’s exciting is that while the tools can assist a healthcare team in identifying potential problems, the process is still driven by a human user, and that makes the process adaptable. And finding a problem feeds back into the process for future patients, as doctors learn more about a pathology that was previously unknown.

Looking more deeply into the volumes of information we collect ultimately helps us shape our understanding of everything from human and natural history to the future of healthcare.

‹ Previous post
Next post ›

Matthew Kyan received a BSc in Computer Science, BEng, and PhD in electrical engineering from the University of Sydney, Australia. He joined Ryerson University in 2008, where he played an instrumental role in establishing the Centre for Interactive Multimedia Information Mining (CIM2). He played several advisory roles in the early stages of Ryerson’s successful incubator – Digital Media Zone (DMZ), and the Centre for Cloud & Context-Aware Computing (RC4). He also led the launch of a new Master’s program in Digital Media, serving as Program Director from 2013-2015.

In 2015, Kyan joined the Lassonde School of Engineering in the Department of Electrical Engineering and Computer Science with a mandate to foster interdisciplinary research in the digital media sector.

Kyan’s core research addresses challenges in the efficient organization, management and analysis of media-rich datasets, while developing more natural and intuitive modes for associated human-computer and computer-mediated interaction. Specializing in audio-visual signal processing and bio-inspired models for learning & recognition, his work finds application in multimedia search and summarization, knowledge-assisted visualization, immersive computing and virtual/mixed realities.

He works with numerous industry partners across the Greater Toronto Area and abroad, from a variety of different disciplines, including: graphic arts & communications, film & cinema, performing arts, biomedical imaging & bioinformatics. His passion revolves around dissolving traditional boundaries that exist in the human-computer interface, and using pattern recognition and computational intelligence to extend/augment the human as they engage in computer mediated experiences.

Kyan won the Siemens National Prize for Innovation, Australia for his work with 3D confocal image analysis. He is a member of the IEEE Signal Processing (SPS), Computational Intelligence (CIS) and Engineering in Medicine and Biology Societies (EMBS), and currently serves as Treasurer for IEEE EMBS, Toronto Chapter.

Research2Reality is a groundbreaking initiative that shines a spotlight on world-class scientists engaged in innovative and leading edge research in Canada. Our video series is continually updated to celebrate the success of researchers who are establishing the new frontiers of science and to share the impact of their discoveries with the public.