Artificial intelligence is opening new possibilities for how we understand and manage lung disease – and can even lead to personalized treatment.
Master’s student Yixiu (Helen) He is immersed in the rapidly evolving applications for machine learning, a subset of AI, at the Advanced Pulmonary Imaging (API) Lab in Robarts Research Institute. A biomedical engineering student pursuing a collaborative specialization in machine learning in health and biomedical science, Helen is among the newest members of the API Lab’s research team.
Helen guides patients as they undergo testing with spirometry, oscillometry and other tests that measure how well their lungs function. The opportunity for direct patient interaction was a pleasant surprise.
Helen He works in the Advanced Pulmonary Imaging (API) Lab. (Colleen MacDonald/Western News)
“I’d thought only medical students could work with patients, but here I’m getting an interdisciplinary experience that combines actual patient visits, along with data analysis and engineering principles to improve health care,” she said.
Helen is learning the technical skills needed to apply AI to health issues, working under the supervision of lab director and Western medical biophysics professor Grace Parraga.
Parraga’s team develops novel non-invasive imaging methods and measurements to identify lung damage and dysfunction not detected using conventional clinical methods, help predict disease worseningand personalize treatments for chronic lung disease.
One of the team’s primary research areas is asthma, a chronic lung condition marked by inflamed, narrowed airways with excess airway mucous. Targeted therapy for asthma patients requires a deeper understanding of pulmonary abnormalities and how they respond to novel therapies.
Parraga, who holds a Tier 1 Canada Research Chair in Lung Imaging, said detecting lung disease in patients with serious symptoms but normal clinical tests is challenging.
“When patients are still breathless, it’s difficult to ascertain who is responding to treatment, and when to try something different,” she said.
Helen’s role is addressing this challenge. In the API Lab, she’s creating tools that will help researchers better understand how patients with severe asthma respond to therapy.
The data Helen collects from the patients’ lung tests is combined with MRI, CT and X-ray scans to reveal subtle differences in inhaled gas distribution known as ‘texture features.’ Not every feature is useful for detecting disease, so a statistical process called Boruta analysis is used to identify the most important features, which are then used to train machine learning models. That’s where AI dramatically shortens the path from data to discovery.
“These models can analyze thousands of pixel-by-pixel comparisons at once to reveal hidden patterns in a tiny fraction of the time it would take a person to do the same by hand,” Helen said.
“It’s amazing to see how much we can learn in such a short period of time with machine learning.” – Helen He, graduate student in biomedical engineering
Using large datasets, machine learning models recognize patterns and make predictions or decisions on new, unseen data. As they learn, they improve without being explicitly programmed for every scenario. Helen is excited by the power of this predictive ability to drive major advances in health care.
“Machine learning really allows for personalized health care,” she said.
Scholarship drives innovative research
Helen is one of several master’s students at Western this year who received a Vector Scholarship in Artificial Intelligence (VSAI) from the Vector Institute, an independent, not-for-profit Canadian research organization dedicated to advancing the machine learning needed to drive innovation and economic growth. As part of its mission to bolster Canadian leadership in AI, Vector provides scholarships to outstanding students entering AI-related master’s programs at Ontario universities.
Parraga says Helen brings solid engineering skills to the team as she helps grow the tools needed to provide tailored therapy to asthma patients.
“This work will improve overall patient outcomes and save time and money,” Parraga said.
Helen is motivated not only by how AI can improve patient care, but how it allows her to quickly gain the knowledge needed for the complexities of interdisciplinary study in the API Lab. Though she didn’t have much coding experience, she used AI tools to learn the programming language Python quickly, allowing her to move on to more difficult tasks.
“It really shortens the learning curve. With AI, you can be the connector – the person who learns different aspects to support interdisciplinary work.”
As an international student from Chongqing, China, Helen said she is grateful for the financial support and networking opportunities provided by the Vector scholarship.
“Vector helps me make connections with industry partners, entrepreneurs and educators. I’ve come to know the Canadian environment, so I can go even further in imaging. I feel this is a really important area for study.”
It’s a field where Canadian researchers shine, Parraga says, noting Canada is at the forefront of medical imaging and pulmonary functional imaging research.
“Western-trained imaging scientists have now established new approaches and strong teams who are positioning novel lung imaging markers and technologies to improve the health of patients worldwide.”
Learn more about how Western is optimizing health for all.
