Session Item

Proton therapy is sensitive to range uncertainties, which mainly originate from the CT-based estimation of the proton stopping power ratio (SPR). Range uncertainties are typically accounted for by margins or robust optimization, where tissue-independent (TI) uncertainties (same uncertainty for all tissues) mainly are used. However, it can be assumed that the range uncertainty margins depend on the specific tissues traversed by the protons. The aim of this study was to investigate the differences between range margins based on tissue-specific (TS) and TI range uncertainties.

The CT-based range uncertainties caused by, e.g., CT image noise, beam hardening, and CT-to-SPR conversion inaccuracies were evaluated for lung, soft and bone tissues to quantify the TS range uncertainties (Fig 1). Proton plans were created using matRad for three patients (pelvic, liver, and head-and-neck (HN)) and a thorax phantom, to evaluate the range uncertainties in different tissue compositions. Conventional optimization (i.e., without robust optimization) was used to isolate the difference between applying the two types of uncertainties. The proton plans were re-calculated after applying range uncertainties that were either TS (different range uncertainties for lung, soft, and bone tissues, as categorized by the CT number of the specific voxel) or fixed TI (0.5%-5.0%).  The re-calculated proton plans were compared based on dose volume histogram (DVH) parameters for the target, ring structures around the target, and organs-at-risk (OARs). For each treatment site, the optimal TI range uncertainty was defined as the one resulting in the largest overlap between the TS and TI DVH values.

The range uncertainties were found to be 8.0% of the proton range for lung, 1.0% for soft, and 2.3% for bone tissues. When comparing the re-calculated proton plans, dose differences were mainly found in the vicinity of the target. It was found that a single TI uncertainty did not provide the best fit for all DVH parameters in any of the treatment sites (Fig 2). Hence, the differences between the two types of re-calculated proton plans were smallest when several TI range uncertainties were used for each treatment site. However, the differences between the DVH values found using TS and TI range uncertainties increased only slightly when using a single TI uncertainty for all DVH parameters for each treatment site. The single best-fitting TI uncertainty was found to be 1.0% for the pelvic and HN patient, 1.5% for the liver patient, and 2.0% for the thorax phantom.

Different range uncertainties were found for lung, soft, and bone tissue indicating that range margins based on TS range uncertainties may be more exact than the standard approach of using TI range uncertainties. A single TI range uncertainty might still be sufficient to capture the TS range uncertainties. However, the value of the TI range uncertainty will be dependent on the treatment site.