What is Proton Therapy
Read our Covid-19 policy

Call Now For More Information

General Questions or
Schedule an Appointment


Current Patients


Physician Line


Read our Covid-19 policy

Artificial Intelligence in Medical Physics

Artificial Intelligence in Medical Physics

There's an exciting development in the treatment planning process for proton therapy: artificial intelligence (AI) software tools that help machines learn to create treatment plans. These tools are already being used clinically for conventional therapy using X-rays at some centers. However, machine- learning-assisted planning has not yet been fully developed for proton therapy. Our Center is one of several proton therapy clinics globally that are actively working with Raysearch Laboratories (developer of our treatment planning software) to further this development.

By feeding hundreds of previously-created proton pencil beam scanning treatment plans into the system, physicists can train these machine learning models to develop treatment plans all on their own.

"We train each model to create a treatment plan for a specific disease site," says Dominic Maes, Senior Medical Physicist at our Center. "The accuracy of the model depends on how many plans we have available for training. Generally, the more plans we enter, the better that model will perform. We have begun building models for prostate and breast cancers and are looking into other disease sites in which we have at least 50 plans available in our treatment planning database, which is the minimum number that Raysearch recommends.”

The aim is to automate certain steps in the planning process, which could further streamline our treatment planning workflow and significantly shorten patients' treatment planning process. At this time, we don’t know how much time we will save on the current one- to two-week turnaround, but the team will evaluate this once the models are tested at the Center. We should add that these tools and models will always be only a step in the process, with human quality oversight of the plans.

"Our team just published our preliminary work on using machine learning-based approaches in proton therapy in the peer-reviewed journal Physica Medica in November 2020," says Maes. "We predict that within the next six months, we will test this functionally for prostate and breast cancer plans."

If it proves useful, our team will explore the clinical use of AI-assisted treatment planning at our Center for actual patients in the near future.

You can learn more details about AI-assisted treatment planning here: https://www.raysearchlabs.com/machine-learning-in-raystation/