Session Item

Water-equivalent path length (WEPL) measurements using flat panel proton radiography (FP-PR) has the potential to enable the detection of proton range uncertainties, the basis of high-precision proton therapy irradiations. Accurate WEPL measurements can be obtained using a FP-PR imaging field with several energy layers, which impose a high imaging dose. In this study we propose a FP-PR method for accurate WEPL determination based on a patient specific imaging field with a reduced number of energies (n) to minimize imaging dose. 

Patient specific FP-PRs of 27x27 cm2 were first simulated across a head and neck phantom (CIRS 731-HN) from a gantry angle of 270° as illustrated in figure 1. An energy selection algorithm estimated spot-wise the lowest energy required to cross the anatomy using a water equivalent thickness (WET) map of the phantom and a FP calibration dataset (figure 1(a)). At each spot coordinate (i,j), subsequent energies in steps of 3 MeV were restricted to certain quantities (n=26, 24, 22, …, 2)(figure 1(a, right)), resulting in a patient specific FP-PR imaging field (figure 1(b)). WEPL maps were reconstructed using the FP calibration (figure 1(c)).

Image quality of phantom WEPL maps obtained with patient specific FP-PRs was assessed via mean absolute WEPL errors (MAE) and standard deviations (SD), against a reference FP-PR with a complete set of energy layers where no spot-wise energy selection was applied.

WEPL accuracy of patient specific FP-PRs was assessed in regions of 4x4 cm2 in the base of skull, neck and brain of the phantom using mean relative WEPL errors (MRE) and SDs with respect to multi-layer ionization chamber PR (MLIC-PR) simulations.

Furthermore, a retrospective dataset of in vivo MLIC-PR acquisitions in three head and neck cancer patients was used to further assess WEPL accuracy of patient specific FP-PR simulations (n=10) employing the energy selection algorithm depicted in figure 1. The imaging fields covered 4x4 cm2 and were delivered/simulated around the treatment isocenter from a gantry angle of 90°. MREs and SDs were computed between patient specific FP-PR simulations and in vivo MLIC-PR measurements. 

The image quality analysis showed MAEs ranging from 2.1±5.1 mm (n=26) to 21.0±16.7 mm (n=2) (figure 2(a)). For n<10, MREs up to 11.4±2.8% were found. For n≥10, MREs were below 0.7±1.6% in the three analyzed anatomical regions of the phantom (figure 2(b)). MREs between simulated patient specific FP-PRs (n=10) and in vivo MLIC-PR acquisitions in the three patients were -0.5±1.5%, -0.4±2.8% and -1.9±1.5%.

A method to accurately measure WEPL using FP-PR with a reduced number of energies tailored to the patient anatomy has been established in silico and evaluated with respect to in vivo patient MLIC-PR measurements in head and neck cancer patients. Patient specific FP-PRs hold the potential to assist online range verification quality control processes within online adaptive proton therapy workflows.