Banafshe Felfeliyan, PhD

Area of Study / Keywords

Computer vision Medical Imaging Machine Learning

About

I am currently a postdoctoral fellow at the University of Alberta. I completed my PhD in the Biomedical Engineering program at the University of Calgary and have both a BSc and MSc in computer engineering from Isfahan University of Technology. I possess experience in research and development related to learning algorithms, computer vision, and medical imaging. I am proud to be a recipient of the Alberta Innovates Postdoctoral Recruitment Fellowship, an esteemed program aimed at fostering innovative and impactful career paths within Alberta's health research and innovation ecosystem. 

I find joy in research! particularly in the areas of learning algorithms, computer vision, imaging, and representation learning, along with their practical applications. I am keenly interested in exploring methods to improve the versatility, dependability, and resilience of AI through diverse approaches.

Research

I am currently working under the guidance of Dr. Jacob Jaremko at the Northern Institute for Deep Learning in Ultrasound (NIDUS) research lab. The primary objective of my postdoctoral project is focused on advancing practical real-world AI applications within the field of medical imaging. This goal revolves around developing and validating deep learning (DL) algorithms tailored for medical image analysis. To ensure progress toward more applicable uses, it's essential to create sturdy and well-calibrated models. Additionally, clinical validation is a crucial step, involving the comparison of the model's performance against that of clinical experts and confirming its correlation with clinical outcomes. In the context of deep learning validation for medical purposes, quantifying the uncertainty linked to individual predictions and calibrating models for equitable results becomes paramount. By leveraging computer vision and deep learning techniques, my emphasis will be on refining and evaluating an automated tool designed to quantify osteoarthritis (OA) in magnetic resonance (MR) images. The ultimate objective is to create a reliable and robust solution that significantly contributes to medical image analysis and enhances patient care.