Monocular 3D Postural Control Monitoring

Balance control largely governs our ability to explore and interact with the world. Aging and neurodegenerative diseases (e.g., Parkinson's disease, multiple sclerosis) can cause deterioration in postural control, which is associated with increased risk of falls, and thus decreased quality of life. Neuromuscular decline results in subtle 3D biomechanical movement patterns, which can currently only be quantified in elaborate whole-room motion capture setups, making naturalistic monitoring outside of the lab challenging. To support balance control assessment in clinical and resource constrained environments, I designed an intelligent monocular imaging system using kinematic priors for quantifying subtle 3D motion from weakly constrained imaging setups. This system was able to distinguish healthy from unhealthy (cerebral hypoperfusion) cases based on sub-millimetre kinematic sway patterns.

Widefield diffuse optical spectroscopic imaging (DOSI) for cancer

Diffuse optics are powerful light-based technologies capable of assessing tissue health (e.g., breast cancer). However, in their traditional form, they are point-based devices, making whole region monitoring difficult. In collaboration with University of California Irvine, we developed widefield diffuse optical spectroscopic imaging (DOSI) for breast cancer screening using computer vision-based AI. Through co-integrative design, we transformed the DOSI measurement system from a single-point single-measurement system to one capable of screening a 10 × 10 cm tissue region for cancer in under 1 min, providing rapid and repeatable biophotonic clinical breast cancer screening based on functional tissue chromophore changes.

Coded Hemodynamic Imaging

Coded hemodynamic imaging (CHI), is a highly novel and portable cardiovascular imaging system that is able to detect blood flow across large areas using specially "coded" light, enabling new ways for both preventive and acute physiological monitoring. This research brings together concepts from biophysical tissue optics, biomedical optics, embedded systems design and image and signal processing to enable first-of-its-kind monitoring. This research has attracted international media attention.

Nominated for Millenium Technology Prize, Technology Academy Finland

Metabolic Oxygen Uptake Prediction

The metabolic rate of oxygen uptake during activity can provide valuable insight into aerobic system function and early pre-clinical indications of degenerative diseases. However, traditional measuring techniques are invasive and prohibit daily monitoring. In collaboration with the Vascular Aging and Brain Health Lab, we developed a physiological regression model to estimate metabolic oxygen uptake using wearable sensor inputs.

Automatic Melanoma Detection

​Melanoma is the most lethal skin cancer, causing 75% of all skin-cancer related deaths, and yet clinical detection is still a very subjective process. In collaboration with our industry and clinical partners Agfa Healthcare and Toronto General Hospital, we developed novel biomedical image processing and machine learning algorithms for skin image correction, feature extraction and classification to create a computer aided diagnostic system.
Proudly powered by Weebly