Fraunhofer Institute for Electronic Nano Systems
One of the most employed ways to treat cancer is through radiotherapy, where the objective is to shrink the tumoral body within the patient through high intensity radiation. Conventional techniques with X-rays are designed to do so with high precision, but they come with the drawback of partially irradiating proximal and distal regions of the tumor, potentially harming healthy tissues and bringing undesired side effects to the patients.
Proton beam therapy has instead the potential to offer a much more conformal radiation dose to patients, as it allows to deliver a high amount of energy in a localized area while sparing regions of tissue around the target. Due to this property, it is therefore of utmost importance to verify that the irradiated area coincides with the tumoral region, as planned and delivered dose can vary due to factors such as patient motion during treatment and uncertainties in the beam range calculation.
NOVO is a project founded by the European Innovation Commission that aims at making real-time dose verification possible during treatment in a non-invasive way. This is achieved by detecting secondary radiations produced by the proton beam within the patient’s body, which can be measured through a newly designed detector and correlated with the actual proton range. As part of the development plan, it will be necessary to design new reconstruction algorithms to map the detector measurements to the delivered dose as well as tools to allow changes in real time to the treatment plan.
The NOVO consortium comprises several universities and institutes with different backgrounds, among which Fraunhofer ENAS is tasked with providing expertise in AI to overcome some of the challenges of the project. In particular, the team will work on developing models for fast surrogate Monte Carlo simulation and for proton range and dose reconstruction starting from detector data. This will require training of deep AI models constrained by the physical and mathematical conditions imposed by the problem.
Find out more on the Novo consortium website and LinkedIn page of the project.