Anatomical modifications occuring during the course of treatment can alter the range of the protons and modify the dose distribution in the target volume and the organs at risk. By linking together multiple open-source software packages, the researchers in adaptive proton therapy have the tools to quantify the divergence between the planned dose and the dose actually delivered to the anatomy of the day by using the in-room CBCT or the in-room CT images. In addition, open source software tools are also available to allow the researcher to quickly develop software prototypes to investigate different strategies for (on-line or off-line) adaptive proton therapy.

In current radiation therapy, patient anatomy is represented by the snapshot of CT images at treatment planning stage. However, during the treatment course, the patient anatomy will change. The changes can results from day to day anatomical variations (inter-fraction organ motion), errors due to patient setup inaccuracy, change in patient condition (weight gain or loss). These deviations from the time of initial simulation to the time of treatment delivery should be accounted for, to ensure compliance of the delivered dose distribution. In addition, because of the finite range of protons, proton therapy is more susceptible to tissue density uncertainty / vairations than photon therapy.

Interesting readings

• Adaptive treatments are a must for protons.
• Adaptation is mandatory for lung IMPT.
• Near real-time automated dose restoration in IMPT to compensate for daily tissue density variations.

The following list describes the different steps of the adaptive PT workflow and gives examples of the software tools available to support the researchers in their work. Combining these open source tools can help you develop more easily software prototypes supporting your clinical research in adaptive proton therapy.

• Robust planning: The initial plan should be as robust as possible to the largest possible number of scenarios. The open source package MIROpt provides the researchers with the tools to make robust optimization of a Pencil Beam Scanning (PBS) plan.
• Patient QA: The initial plan must be QA-checked (patient QA). In order to reduce the workload of the medical physycists, a research project investigates the possibility of using the machine logs recorded during a dry run of the delivery of the plan by the treatment machine. The machine logs can be analysed using tools developed in OpenREGGUI.
• In-room CBCT: The anatomy on the day of treatment is monitored by Cone Beam Computed Tomography (CBCT) installed on the treatment machine. As the quality of CBCT is not sufficient to accurately estimate the dose delivered by the proton beam, a virtual-CT. is computed first by deforming the planning CT onto the CBCT or the day. OpenREGGUI provides the tools to perform the virtual CT computation.
• Clinical indicators: Using the in-room CT or the virtual-CT, different "clinical indicators" can be estimated, based either on the dose distribtuion (DVH, D95,...) or on the Water Equivalent Pathlength (WEPL). A research project identifies the relevant clinical indicators to use to trigger a replanning. OpenREGGUI and the CAPTAIN platform provide a web interface to automate the computation and review of the clinical indicators by the researchers in a web interface.
• Dose restoration: The MIROpt software provides a way of restoring the dose distribution of the initial plan to the antomy of the day. This is the concept of dose restoration.
• Contour propagation: In order to achieve a full plan re-optimization, it is necessary to propagate the anatomical contours (defined on the planning CT) to the CT or virtual-CT of the day. This requires a reliable contour propagation algorithm. A research project investigates the possibility of using the open-source deep learning library TensorFlow for this purpose.
• Pre-treatment patient QA: Before the new plan is delivered, a patient QA of the adapted plan is required. In case of on-line plan adaptation, it is not possible for the physicist to make a measurement inside the treatment room. A research project investigates the possibility of using the open MCsquare fast open source dose engine in order to run an independent secondary dose engine check of the adapted treatment plan computed by the TPS.
• Post-treatment QA: Following the delivery of the treatment plan, the machine logs can be used to run a patient QA. The OpenREGGUI platform provides several tools for the analysis of machine logs.