Developing and Deploying an Automated Data Analysis Infrastructure for Brachytherapy | UHN Medical Physics Seminar

Brachytherapy plays a critical role in the management of locally advanced cancer, and recent advances in MRI-guided brachytherapy have led to improvements in survival compared to historical controls, especially for bulkier tumours. The increasing complexity and number of fractions in adaptive brachytherapy treatments lead to the development of automated tools for extracting doses for both quality assurance and research to improve clinical care.

Dosimetric analysis for Brachytherapy requires the integration of treatment planning across all brachytherapy fractions and plans. However currently, in both clinical and research practice, the full course delivery is assessed by comparison of dose-volume histogram (DVH) parameters for each specific fraction rather than a full spatial analysis of a 3D cumulative dose. Furthermore, brachytherapy population data analysis requires manual extraction of dose using the current treatment planning systems and is very cumbersome and inefficient for a large number of patients. Despite having plenty of clinical data to work with, the lack of infrastructure hinders our ability to fully process and analyze this data to answer important clinical questions.

Thus, there is a general need, across various projects and patient cohorts or disease sites, for large-scale dosimetric analysis, notably the standalone calculation of DVHs. MIRA® is a full-stack radiotherapy analytics platform developed at the Princess Margaret Cancer Centre that integrates with systems at the University Health Network Radiation Medicine Program (UHN RMP) to build automated pipelines for image processing and data analysis.

Please find the slides here.

Over 60 medical physicists, graduate students, residents, practicing radiation oncologists and therapists were in attendance. The seminar seminar was also live-streamed to all campuses of UHN partner hospitals.