We use cookies to help provide you with the best possible online experience. Learn more


Machine learning for intelligent treatment planning

Cancer treatment represents one of the most exciting applications of data analytics and machine learning technologies. Machine learning already supports the identification of diseases and diagnosis, through to treatment planning and aftercare. By analyzing thousands of different data points, advanced algorithms in RayStation* can help clinics save time and increase consistency by automating plan generation and organ segmentation.

*Subject to regulatory clearance in some markets.

Key features

  • Deep learning based automated segmentation of CT structures
  • Deep learning based automated treatment plan generation
  • Validated models are released and upgraded several times per year
  • Spend less time on repetitive tasks
  • Get more time for patient consultations and complex cases

Deep Learning Segmentation now included in RayStation 11B and beyond

RaySearch is continuously improving the released models and is planning to release new or updated Deep Learning* Segmentation (DLS) models regularly going forward, independent of version releases.

Deep learning capabilities in RayStation help make image segmentation quicker and more consistent. A high-speed GPU-powered algorithm is capable of producing consistent segmentation results using segmentation models that have been trained and evaluated on clinical data for different body sites. A deep learning model can segment multiple structures in less than one minute.

Deep learning segmentation in RayStation

In this webinar you learn more about how deep learning segmentation can be implemented at your clinic. Our senior application specialist Grietsje Schregardus-Abma will demonstrate the latest release of segmentation models and talk about how to go live with DLS. Sigrun Saur, St Olavs Hospital, will talk about training, validation and clinical implementation of a DLS model for breast cancer, a joint venture together with RaySearch.


The world’s first machine learning plan generation module

RaySearch partnered with Princess Margaret Cancer Center in Canada to develop the world’s first machine learning treatment plan generation module. In May 2019, patients with localized prostate cancer were treated using machine learning treatment plans generated in RayStation as part of a compensative evaluation study. See the study here

RaySearch received 510(k) clearance from the U.S. Food and Drug Administration for RayStation 8B , which was the first machine learning applications in a treatment planning system on the radiation oncology market today.

Clinics can get personalized treatment plans in RayStation to benefit from the experience of one of the world’s leading cancer centers, generated in minutes by selecting a pre-trained machine learning model. One or multiple deliverable treatment plans can be automatically generated with varying target/OAR tradeoffs.

WEBINAR: Deep learning planning in RayStation

In this webinar we are presenting our latest release of machine learning planning models, machine learning news in RayStation 11B and how machine learning planning can be implemented at your clinic. Demonstrating how your clinic can configure and commission a pre-trained model to your protocol, planning trade-offs and treatment machines.



Physics World met up with Fredrik Löfman, Head of Machine Learning at RaySearch, at 2019 ASTRO Annual Meeting in Chicago to discuss how machine learning capabilities in RayStation can help revolutionize treatment planning.


Testimonial from Tom Purdie, PMH

“Machine learning is a natural fit for automating the complex treatment planning process. We expect it will enable us to generate highly personalized radiation treatment plans more efficiently, thereby allowing clinical resources or specialist technical staff to dedicate more time to patient care. Deemed clinically acceptable by experts around the world, RayStation algorithms generate high-quality treatment plans that are preferred or deemed equivalent to clinical plans.”

Tom Purdie
Medical Physicist, Princess Margaret Cancer Center