Radiotherapy treatment planning is a heavily computerized process relying on accurate dose calculation and optimization to enable control of cancerous tissue with high dose while sparing the healthy tissue and critical organs.
The Radiotherapy Optimization Group researches the integration of physical, biological and mathematical models into the treatment planning process for radiation therapy. The focus lies on pre-clinical research and prototyping new approaches in dose calculation or dose optimization.
Currently, the key areas of research are:
Treatment planning “beyond dose”:
Conventional treatment planning prescribes dose (i.e., the energy absorbed by the tissue) to define treatment goals and toxicity endpoints. However, the biological response does also depend strongly on the radiation quality, particular for charged particle therapy, tissue types and biochemical environment. Thus, we are primarily interested in using other physical quantities with stronger response correlation (e.g., from nanodosimetry) for treatment planning and the use of artificial intelligence to directly plan on predicted outcome.
Combined treatments:
Locations like Heidelberg have access to many different irradiation modalities like photon, proton, and ion treatments, but in general, their individual availability varies hugely. Combining those modalities within the treatment regime can improve resource usage, but is far from trivial due to different biological effectiveness. We are interested in finding the optimal combination and planning strategies for such combined treatments
Integration of Artificial Intelligence:
The integration of AI into the treatment planning workflows shows to major potentials: They can either streamline otherwise tedious human tasks (e.g., organ segmentation or plan tuning) or accelerate otherwise computationally demanding numerical computations like Monte Carlo simulation ore optimization algorithms. Our primary focus here lies on AI-based dose calculation and using highly complex AI treatment outcome prediction models directly in plan optimization.
In all of our research we advocate for open science. Much of our research is executed within our open-source treatment planning toolkits matRad (www.matRad.org) and pyRadPlan.
