Banafshe Felfeliyan

Ph.D. | Computer Vision | Machine Learning | Deep Learning | Biomedical Engineering | Medical Imaging

Toronto, Ontario, Canada

About

As a Ph.D., my commitment lies in the pursuit of knowledge across a broad spectrum of ideas and insights within my field! I am a self-motivated and ambitious researcher with 6+ years of research and development experience in learning algorithms, computer vision, medical imaging, and their real-world impact. It is exciting for me to explore solutions to address the generalizability, reliability, and robustness of machine learning solutions as these features are vital in real-world applications. My interests in both computer engineering and medicine led me to get involved in biomedical-based research. In my Ph.D., I quantified osteoarthritis (OA) features from knee and hip MRI scans using deep learning to investigate OA progression, by proposing self-supervised and weakly supervised learning methods. I have the following experience and skills: - Strong Machine Learning & Deep Learning foundation knowledge - Worked with the mainstream DL architectures - Worked with Machine Learning frameworks TensorFlow, Keras, scikit-learn ... - Computer vision and Image processing - Multidisciplinary collaboration experience

Experience

  • Member of Technical Staff at AMD
    Feb 2026 - Present · 5 mos

  • Post Doctoral Fellow - Oncology - Machine Learning at AstraZeneca
    Jan 2025 - Jan 2026 · 1 yr 1 mo

  • Postdoctoral Researcher (Radiology & Diagnostic Imaging Dept.) at University of Alberta
    Apr 2023 - Jan 2025 · 1 yr 10 mos

    This postdoctoral project focuses on advancing practical applications in medical imaging through the development and validation of a specialized deep learning (DL) algorithm for medical image analysis. Creating robust and well-calibrated models is vital to enhancing applicability. Additionally, clinical validation—where the model's performance is compared with that of clinical experts and its correlation with clinical outcomes is confirmed—plays a pivotal role. In the context of validating deep learning for medical applications, it becomes crucial to quantify uncertainty in individual predictions and calibrate models for equitable outcomes. Leveraging representation learning techniques, I will concentrate on refining and evaluating an automated tool designed to quantify osteoarthritis (OA) in magnetic resonance images (MRIs). The ultimate aim is to create a dependable solution that greatly improves medical image analysis and patient care. My project, titled "Automated AI MRI Biomarker Profile for Osteoarthritis," addresses the urgent need for more precise OA diagnosis and assessment. Given OA's high prevalence in the world and the associated challenges, a standardized tool for quantifying structural damage and active disease in OA is imperative. I seek to refine and assess an automated OA quantification tool, with the intention of widespread use in the numerous MRIs conducted annually in Alberta.

  • University of Calgary (Calgary, Alberta, Canada)
    • Doctor of Philosophy (Ph.D.)
      Sep 2018 - Mar 2023 · 4 yrs 7 mos

      I am developing deep learning algorithms (DL) for analyzing MRI scans of osteoarthritis patients for my Ph.D. project. My project includes a variety of Machine learning and image processing tasks including segmentation, object detection, classification, & data analysis to relate the Osteoarthritis features & pathologies to clinical scores. I addressed the limitations of previous DL methods including low generalizability (by domain adaptation) & the need for the pixel-wise label (proposed a soft training & weak label training), & limited annotated data (by self-supervised learning).

    • Graduate Teaching Assistant
      Sep 2018 - Dec 2021 · 3 yrs 4 mos

      I have been a teaching assistant for a variety of courses, including: - Artificial Intelligence & Machine Learning (ENSF 411), 2021 - Machine Learning (ENSF 611 & ENSF 619) 2019 - 2020 - Computer programming (ENCM 335), 2019 - 2021 - Human motor control system and learning (KNES 251), 2018 Tasks - Supervised laboratory and tutoring students - Provided constructive feedback - Developed laboratory instructions (ENSF 619) - Evaluated Student’s productivity

  • Machine Learning Engineer at McCaig Institute for Bone and Joint Health
    Sep 2017 - Aug 2018 · 1 yr

    MOJO Motion Assessment Laboratory Center for Mobility and Joint Health Project: Bone Segmentation in MRI images using deep learning Description: - Research and Development (R&D): deep-learning, AI, and medical image analysis - Developing deep-learning algorithm for bone segmentation for data processing pipeline for the High-Speed Bi-planar Videoradiography (HSBV) system - Presentation of the developed algorithm at peer-reviewed conferences