Encrypted login | home

Program Information

On the Potential of Dual-Energy CT to Predict Nuclear Interaction Cross Sections and Radiation Length of Therapeutic Proton Beams

no image available
E Baer

E Baer1,2*, A Lalonde3 , G Royle1 , H Bouchard3 , (1) University College London, London, UK, (2) National Physical Laboratory (NPL), Teddington, UK, (3) Universite de Montreal, Montreal, Quebec


SU-I-GPD-J-62 (Sunday, July 30, 2017) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose: Recent studies have demonstrated improvements in predicting proton stopping powers (SP) using dual-energy CT (DECT). In the context of proton treatment planning, additional material specific information other than SP, such as nuclear cross sections (NCS), radiation length (or scattering power) and electromagnetic interaction mean free path (MFP) is useful to produce accurate dose distributions in patient geometries. This work aims at showing that DECT can benefit model-based calculation techniques accounting for these effects.

Methods: CT numbers of 70 reference tissues are simulated for two spectra (80 kVp, 140Sn kVp). Elemental compositions are extracted using the eigentissue decomposition (ETD) method (Lalonde and Bouchard, 2016). NCS of reference tissues are calculated from elemental NCS, as tabulated in ICRU report 63, using the resulting ETD elemental compositions. Radiation lengths are estimated from elemental radiation lengths weighted by elemental compositions, and electromagnetic MFP are calculated from the total ionization cross section. We compare the ETD-predicted values to reference values and benchmark our technique against gold standard SECT predicted data.

Results: DECT predicts NCS well within 1% for most of the 70 tissues, with only few exceptions. The use of DECT-extracted elemental compositions for NCS outperforms the bi-linear SECT calibration curve, especially for high energies, with root mean square (RMS) errors of 0.46% and 1.05% at 250 MeV over all tissues for DECT and SECT respectively. Similar findings are reported for the radiation length, with RMS errors of 0.95% and 2.00%, respectively, as well as a MFP accuracy of 0.07% and 0.43%.

Conclusion: DECT predicts material-specific NCS, radiation length and MFP accurately and outperforms state-of-the-art SECT calibrations. Accurate knowledge of these quantities can improve model-based treatment planning. Potential applications include lateral scaling of pencil beams, estimation of the nuclear halo and the definition of input parameters for Monte Carlo simulations.

Contact Email: