openPR - A computational tool for CT conversion assessment with proton radiography.

Area of Science:

• Medical Physics
• Radiotherapy
• Image Processing

Background:

• Proton therapy planning relies on converting CT Hounsfield Units to proton stopping powers, a key source of uncertainty.
• Proton radiography offers range error maps, but these can be influenced by multiple error sources beyond CT conversion.

Purpose of the Study:

• To develop and validate a workflow for assessing CT conversion uncertainty using only proton radiographs.
• To quantify the specific contribution of CT conversion to range uncertainty in proton therapy.

Main Methods:

• A novel workflow was implemented using proton radiographs from a multilayer ionization chamber to assess CT conversion.
• The workflow involved CT/radiograph registration, sample-specific model validation, and water-equivalent path length (WEPL) error estimation.
• Applied to a pig head for validating the CT calibration at the PARTICLE proton therapy center.

Main Results:

• The workflow showed an overall overestimation of CT conversion uncertainty by 71% on a pig head compared to standard safety margins.
• Significant underestimation of range uncertainty occurred in areas with cavities, correlating with low Hounsfield Units.
• CT calibration was adapted by widening the air HU interval to -950 to mitigate these localized errors.

Conclusions:

• The proposed workflow effectively validates CT conversion to relative proton stopping powers, revealing an overall uncertainty overestimation.
• Cavities were identified as a major source of localized range errors, suggesting potential for reducing treatment planning uncertainties.
• Implementation in the open-source tool openPR facilitates further studies on patients and diverse tissues to refine uncertainty estimations.