by John R. Fischer
, Senior Reporter | December 24, 2018
Software provider RaySearch has released RayStation 8B, the latest version of its treatment planning system and the first of any TPS on the market to offer machine learning and deep learning applications, according to the company.
Offering automated treatment planning and organ segmentation, the system is expected to boost efficiency and consistency in clinics and hospitals.
“The features include state of the art neural network architecture, and there are significant differences, compared to existing automation,” Johan Löf, CEO of RaySearch, said in a statement. “They are faster, may generalize better, and it is easy to share machine learning models. The two features work well together and are ideal for adaptive workflows as they are both fast and consistent.”
With machine learning and deep learning, users of the RayStation 8B can train models on the available data within a clinic, or rely on pre-trained models developed by RaySearch.
The platform also offers a module for evaluating the robustness of treatment plans, particularly the influence leveraged by uncertainties in patient setup and density interpretation of CTs. Clinicians can easily create and assess multiple scenarios with different uncertainty settings simultaneously and with quick decision support, using different robustness metrics connected to the clinical goals of the treatment.
Another component is the Monte Carlo dose engine, which provides improvements in photon planning and optimization. The algorithm of the engine improves not just accuracy but uses the GPU to enable fast dose computation, such as in a dual arc VMAT plan, which can be computed in less than 60 seconds. This is at least one order of magnitude faster than any other system on the market.
Also new is planning for boron neutron capture therapy, a type of radiation therapy that enables cancers to be targeted at the cellular level. The feature was developed in collaboration with Sumitomo Heavy Industries and Neutron Therapeutics.
“It is going to be very interesting to see how far ahead our users can advance with these new powerful tools,” said Löf.
Other innovations include support for directly deliverable MCO for VMAT; collimation of individual energy layers with the Adaptive Aperture of the Mevion S250i HYPERSCAN proton therapy system; and the handling of relative biological effectiveness for proton dose.