Speaker
Description
Artificial intelligence (AI) is reshaping medical physics by enabling more automated, quantitative, and patient-specific cancer care. My research program develops machine learning methods to improve cancer diagnosis, radiation treatment planning, dose calculation, and outcome prediction using multimodal patient data, including medical imaging, digital pathology, molecular information, and clinical text. Another branch of my research is the collection and curation of large, multi-institutional datasets to support robust, clinically generalizable AI models.
This presentation highlights how AI is used to model, optimize, and personalize radiation-based cancer treatments. One area of focus is brachytherapy, a form of radiation therapy in which radioactive sources are placed close to or inside the tumour. In this context, I will describe AI-based treatment planning methods that optimize source placement and treatment delivery, with the goal of maximizing tumour dose while minimizing exposure to surrounding healthy tissue. I will also present deep learning tools for rapid dose estimation that approximate computationally intensive physics-based Monte Carlo simulations while preserving clinical accuracy. Another component of the presentation focuses on data collection and curation efforts, including the assembly of large, multi-institutional and multimodal datasets needed to train and validate clinically generalizable AI models. Finally, I will discuss multimodal prediction models that combine imaging, pathology, molecular, and clinical data to improve understanding of treatment response, progression-free survival, and individualized patient risk.