Skip to content

Research

Developing robust uncertainty-aware machine learning methods for medical imaging applications

Research Interests

Machine Learning

Deep learning, ensemble methods, uncertainty quantification

Probabilistic Modeling and Forecasting

Bayesian approaches, statistical modeling

Uncertainty Estimation

Model uncertainty, aleatoric and epistemic uncertainty

Generative Models

Focusing on Diffusion Models and Flow Matching techniques

Medical Imaging

Echocardiography, ultrasound imaging, cardiac analysis

Applied Mathematics

Optimization, signal processing, matrix analysis

Current Research

PhD Research (2021-Present)

University of British Columbia, Robotics and Control Laboratory

My current research focuses on developing robust uncertainty-aware machine learning methods for medical imaging applications, particularly in echocardiography and cardiac function assessment. I work on creating AI systems that can reliably quantify their own confidence, making them suitable for deployment in clinical environments.

Master's Research (2018-2021)

Thesis: "Towards a robust estimation of ejection fraction: a deep uncertainty aware approach"

Developed novel deep learning frameworks incorporating Bayesian uncertainty estimation for more reliable cardiac function assessment from echocardiographic videos. This work laid the foundation for my current PhD research.

Selected Projects

Asymmetric Observer Variability in Echocardiography Quality Assessment

Bayesian Decision Theoretic False Negative Reduction with Skewed Posteriors - addressing the challenge of asymmetric costs in medical diagnosis.

Unsupervised Topologically Deformable Probabilistic Registration of Echocardiograms

Novel approach to cardiac image registration using topological constraints and probabilistic methods for better image alignment.

Medical Image Registration Survey

Comprehensive review of medical image registration methods, from classical mathematical approaches to modern machine learning techniques.

Internship Experience

Summer 2017

Medical Image and Signal Processing Research Center, Isfahan, Iran

Project: "Automatic Diabetic Retinopathy Detection and Fundus Photograph Segmentation using Deep Learning and Dictionary Learning"

Developed automated systems for early detection of diabetic retinopathy from retinal images, combining deep learning and traditional computer vision techniques for improved diagnostic accuracy.