ESTRO 38 Abstract book

S1122 ESTRO 38

for the stoichiometric method. This study investigated whether the SPR accuracy of the stoichiometric method could be improved by applying other CT number

estimation methods. Material and Methods

Six published CT number estimation methods (Table 1) were investigated, including the method originally proposed for the stoichiometric method. All methods proposed a calibration method to characterize the x-ray energy spectrum based on CT scans of an electron density phantom, without the need for direct measurements of the x-ray spectrum. Therefore, these methods can all be used in the first steps of the stoichiometric method to estimate the CT numbers for reference human tissues, which in the last step is used to fit an empirical piecewise linear conversion curve from CT numbers to SPRs. CT scans of a tissue equivalent phantom (Cone-Beam Electron Density Phantom, Gammex, Middleton, WI, Fig. 1) were acquired at 120 kVp with a Flash scanner (Siemens Healthineers, Forchheim, Germany). The CT number estimation accuracy of the six methods was tested in a fully self-calibrated evaluation for the twelve insert materials used in the calibrations (measured vs estimated CT numbers), as well as theoretically for the reference human tissues (theoretical vs estimated CT numbers). Stoichiometric conversion curves were fitted based on each set of estimated CT numbers for the reference human tissues, applying the same curve fitting recipe for all six conversion curves (Fig. 2). The SPR accuracy was evaluated for published proton range measurements of organic beef and pork tissue samples, which had been scanned with the same CT scanner.

Conclusion Highly-accelerated 3D-MRI could enhance positional verification efficiency in the HN MRgRT without compromising guidance accuracy. EP-2044 CT number estimation techniques for the stoichiometric method to predict proton stopping power V. Taasti 1 1 Memorial Sloan Kettering Cancer Center, Medical Physics, New York, USA Purpose or Objective The stoichiometric method is a well-established method, often used in proton therapy planning to estimate the stopping power ratio (SPR) of a patient’s tissue. This method includes a CT scanner characterization and a subsequent CT number estimation for reference human data from the literature. The method originally proposed for these steps is based on a parametrization of the x-ray attenuation coefficient with fixed parameters for the target material dependency. However, other CT number estimation methods exist, which can equally well be used

Results The CT number estimation method originally proposed for the stoichiometric method provided the least accurate CT number estimates with a root-mean-square error (RMSE) of 3.1% (Table 2). The lowest RMSE was 0.2%; this was obtained with a method based on two effective energies to characterize the CT scanner energy spectrum. In the theoretical evaluation, the deviations for the original method were found to increase with density for bone tissues (Fig. 3). The SPR accuracy for the organic tissues varied between the methods, with RMSEs between 1.3% and 0.5% (0.7% for the original method).