Automated
Treatment
Planning

Automated breast planning

The automated breast planning solution in RayStation is the first step in our journey to automate routine and standard procedures. RayAutoBreast was initially developed in Toronto, Canada. Between 2009 and 2012, clinicians at Princess Margaret Hospital (PMH) ran a large-scale clinical study to evaluate the performance of automated treatment planning for intensity modulated radiation therapy (IMRT) for tangential breast treatments. They concluded that it could add tremendous degrees of efficiency and standardization, as well as quality to current treatment planning processes. The use of automated procedures will allow for faster adoption of IMRT together with increased access to care improvements for breast cancer patients [1].

RayAutoBreast provides tools for automated generation of tangential breast IMRT plans using heuristic optimization and includes features such as:

• Detection of radio-opaque markers defining the breast.
• Contouring of all the relevant target and risk organs.
• Setup of beams, including heuristic optimization of gantry and collimator angles.
• Creation of objective functions, optimization and segmentation settings and clinical goals.

[1] T.G. Purdie et al., “Automated Planning of Tangential Breast Intensity-Modulated Radiotherapy Using Heuristic Optimization”, Int.J. Radiation Oncology Biol.Phys.Vol. 81, No.2, pp. 575-583, 2011.

Fallback planning

Fallback planning is a tool for creating additional plans to be used in a contingency situation, enabling a patient to be treated on another machine, possibly with a different modality and/or treatment technique, in case the original machine is unavailable. It can drastically reduce planning time in emergency situations when a machine is down allowing the patients´ treatment to continue and reducing stress on staff.

• Fallback planning uses a dose mimicking function to replicate the DVHs of a given plan, but for a different machine or treatment modality.
• Plans of any modality, including proton and tomotherapy plans, can be replicated using photon plans like 3D-CRT, IMRT, or VMAT.
• Fallback plans are automatically generated after plan approval based on previously created protocols.
• No user interaction is required as this is a fully automated plan creation process. If needed the created fallback plans can be manually modified after the automatic generation.
• Fallback plans can be compared and evaluated using a number of visual tools (DVH curves, dose differences, etc.).
• A Fallback plan can be approved and used for delivery in future fractions. It is also possible to convert back to the original plan.
• With dose summation tools, two plans can be combined using their delivered fractions so that actual composite dose can be visualized on the patient data set.

Plan Explorer

Plan explorer is based on the capability to automatically generate a large number of treatment plans for defined clinical goals and combinations of treatment techniques and machines. It also provides efficient means to filter and browse among plan candidates to find the most desired one. In addition, the plan candidates can be generated using machine learning models which also enables automatically generating several plans from one model with different tradeoff priorities.

Plan Explorer brings many potential clinical benefits and now you can:

• explore more of the solution space to ensure that every radiation treatment is delivered with the highest possible efficiency, with an optimal combination of treatment technique and machine;

• maximize the use of your current treatment delivery machines;

• generate plan candidates using machine learning models;

• explore treatment plan tradeoffs;

• and get more time to evaluate the plans

Webinar: Plan Explorer redefines automated planning

Part 1: Freddie Cardel outlines the concept of automatic plan generation with Plan Explorer and demonstrates a completely new level of automation. He explains the approach, where large numbers of high-quality treatment plans are automatically generated for defined clinical goals and combinations of treatment techniques and machines and shows how these plans can be easily filtered and browsed to find the most suitable candidates to be evaluated.

Part 2: Erik Korevaar and Roel Kierkels present the first findings of the clinical evaluation they performed at University Medical Center Groningen, the Netherlands, and describe how the center envisions the use of the tool in clinical practice.