Research summary
Digital technologies are transforming health care tools, such as point-of-care ultrasound. Advances in machine learning offer enormous potential to improve diagnostic accuracy and outcomes for patients with help from these tools, but challenges remain: while portable ultrasound machines are widely available, there is a shortage of trained users, particularly in rural and remote areas. Machine learning can help—but only if it is trained on large, diverse datasets.
Dr. Purang Abolmaesumi, Canada Research Chair in Biomedical Engineering, is developing a robust machine learning framework that can adapt to users with varying levels of expertise and recognize unfamiliar inputs. He and his research team are advancing self-supervised learning, improving accuracy in noisy data environments, and estimating uncertainty in predictions. They are also designing artificial intelligence systems that can help interpret ultrasound images. Together, these innovations aim to support broader, more equitable access to high-quality diagnostic imaging in Canada and beyond.