Research
Developing robust uncertainty-aware machine learning methods for reliable medical imaging applications.
PhD Research
University of British Columbia, Robotics and Control Laboratory
My research addresses the critical challenge of reliability in medical AI. I develop uncertainty-aware deep learning frameworks that can quantify their own confidence. Specifically, I focus on echocardiography analysis, where assessing image quality and model certainty is crucial for clinical deployment. By bridging Bayesian methods with modern deep learning, I aim to create systems that clinicians can trust.
Core Interests
Uncertainty Estimation
Aleatoric & epistemic uncertainty quantification
Medical Imaging
Cardiac ultrasound & automated diagnosis
Generative Models
Diffusion models & Flow Matching
Robustness
Distribution shift detection & OOD generalization
Deep Learning
Neural architectures & optimization
Applied Math
Statistical modeling & signal processing
Prior Research
Master's Thesis
"Towards a robust estimation of ejection fraction: a deep uncertainty aware approach"
Pioneered the use of Bayesian uncertainty estimation in echocardiogram video analysis. Demonstrated that modeling uncertainty significantly improves robustness against noisy clinical data.
Diabetic Retinopathy Detection
Research Intern, Medical Image and Signal Processing Research Center
Developed hybrid algorithms combining dictionary learning and deep CNNs for automated detection of diabetic retinopathy from fundus photography.