image: SFU computing science professor Ghassan Hamarneh is using his medical imaging analysis expertise to help UBC researchers understand the role of caveolin-1 (CAV1) in certain cancer types. view more
Credit: SFU
SFU computing science professor Ghassan Hamarneh is using his medical imaging analysis expertise to help UBC researchers understand the role of caveolin-1 (CAV1) in certain cancer types.
CAV1 is a protein associated with poor outcomes in aggressive breast and prostate cancer. It is also associated with tumor metastasis and tumor suppression, but it is unclear how CAV1 differentiates between these two roles.
UBC biologist Ivan Robert Nabi and his team are investigating the contribution of CAV1. To do so, they use machine learning-based analyses of super-resolution microscopy to study and visualize caveolae and scaffolds - tiny structures that are found within a cell, and in which CAV1 is found.
Using a graph-based approach, Hamarneh and the other researchers are able to capture and process three-dimensional coordinates of data, known as point clouds, with nanometer resolution to see cellular processes.
Given the complexity and volume of the data, filtering through all this data manually is inefficient. For this reason, the researchers are leveraging computer vision and artificial intelligence (AI) methods to automatically analyze the data. This is done through classical machine learning approaches where the researchers train the machines to examine specific features and feed the data into an algorithm. They also use deep learning approaches where the researchers do not impose their own bias in terms of what features they are looking to examine and have the machine discover which features stand out.
"With deep learning, the machine is discovering features for you, but sometimes these features are not easily interpretable," says Hamarneh.
"We are examining interpretable AI methods, where we look at what the AI discovered and expose it for biologists to gain insight."
Since beginning their collaboration a few years ago, Hamarneh and Nabi have combined their expertise to develop software tools to make this process possible. In collaboration with SFU postdoctoral researcher Ismail Khater, they developed Super Resolution Network Analysis, which helps the researchers analyze biological clusters and can be applied to different biological questions.
Their research efforts have not gone unnoticed. Recently, the research groups received $910,000 from Canadian Institutes of Health Research (CIHR) for the next five years.
"We're very pleased with this funding because it allows us to continue this work and, most importantly, provide salaries for graduate students and postdocs to help us with our research," says Hamarneh.
"To receive very positive feedback on our research is encouraging because it gives validation to the work that we are doing."
While the researchers are focused on understanding CAV1's role in breast and prostate cancer, they are also interested in how their current research can be beneficial to other areas of biology moving forward.
"Our long-term goal is to facilitate biological discovery, for the betterment of human health, by bridging the gap between computational tools and biological questions," says Hamarneh.
###
*The collaboration between Hamarneh and Nabi's labs involved several graduate students and postdoctorates who were instrumental in this interdisciplinary research.