Project 460077

Patients undergoing radiotherapy have daily images to verify their positioning before treatment delivery. As can be routinely observed in the clinic, these images often demonstrate important anatomical changes that are visible to the naked eye. With some exceptions, these changes are often ignored, as their clinical relevance is largely unknown, or speculative at best. Computers now allow us to analyze these big data sets to search for hidden multi-parametric biomarkers buried in the image features that would have the potential to predict how patients will respond to their treatment. In this work, we expect to create a predictive tool, a stand-alone software for automated radiomics integration, that could monitor patients over the course of their radiotherapy, and use this information to adapt the course of treatment, improve outcomes and reduce toxicity. We will build an infrastructure that combines data from different sources (electronic medical chart, radiotherapy plans, daily images, biopsies, lab work, etc.) for patients treated for head and neck cancer or lung cancer at the Centre Hospitalier de l'Université de Montréal. The infrastructure and processes will allow doctors, physicists, data experts, patients and industrials to communicate and work together towards an efficient and accurate predictive tool.