SU-C-204-04: Patient Specific Proton Stopping Powers Estimation by Combining Proton Radiography and Prior-Knowledge X-Ray CT Information

Authors

  • Collins-Fekete CA,

    1. Massachussetts General Hospital, Quebec, Quebec
    2. Ion Beam Application, Louvain-la-neuve, Belgium
    3. Aarhus University, Aarhus, Aarhus
    4. Centre Hospitalier Univ de Quebec, Quebec, QC
    5. Mass General Hospital; Harvard Medical, Boston, MA
    Search for more papers by this author
  • Brousmiche S,

    1. Massachussetts General Hospital, Quebec, Quebec
    2. Ion Beam Application, Louvain-la-neuve, Belgium
    3. Aarhus University, Aarhus, Aarhus
    4. Centre Hospitalier Univ de Quebec, Quebec, QC
    5. Mass General Hospital; Harvard Medical, Boston, MA
    Search for more papers by this author
  • Hansen D,

    1. Massachussetts General Hospital, Quebec, Quebec
    2. Ion Beam Application, Louvain-la-neuve, Belgium
    3. Aarhus University, Aarhus, Aarhus
    4. Centre Hospitalier Univ de Quebec, Quebec, QC
    5. Mass General Hospital; Harvard Medical, Boston, MA
    Search for more papers by this author
  • Beaulieu L,

    1. Massachussetts General Hospital, Quebec, Quebec
    2. Ion Beam Application, Louvain-la-neuve, Belgium
    3. Aarhus University, Aarhus, Aarhus
    4. Centre Hospitalier Univ de Quebec, Quebec, QC
    5. Mass General Hospital; Harvard Medical, Boston, MA
    Search for more papers by this author
  • Seco J

    1. Massachussetts General Hospital, Quebec, Quebec
    2. Ion Beam Application, Louvain-la-neuve, Belgium
    3. Aarhus University, Aarhus, Aarhus
    4. Centre Hospitalier Univ de Quebec, Quebec, QC
    5. Mass General Hospital; Harvard Medical, Boston, MA
    Search for more papers by this author

Abstract

Purpose:

The material relative stopping power (RSP) uncertainty is the highest contributor to the range uncertainty in proton therapy. The purpose of this work is to develop a robust and systematic method that yields accurate, patient specific, RSP by combining 1) pre-treatment x-ray CT and 2) daily proton radiograph of the patient.

Methods:

The method is formulated as a linear least-square optimization problem (min||Ax-B||2). The parameter A represents the pathlength crossed by the proton in each material. The RSPs for the materials (water equivalent thickness (WET)/physical thickness) are denoted by x. B is the proton radiograph expressed as WET crossed. The problem is minimized using a convex-conic optimization algorithm with xi<xi+1 constraint. The Gammex RMI-465 phantom, with 13 material inserts, is used to investigate the optimal RSPs. The proton radiograph (B) is produced using Geant4 simulation. The matrix A is computed by calculating proton trajectories through the CT with the cubic-spline approach. Geant4 RSPs are used as reference and the clinical HU-RSP calibration curve is used as comparison. The convergence is investigated as a function of the number of angles and protons used. The radiographs angle which yield the most accurate RSPs are investigated.

Results:

Optimization with 9 angles and 104 protons/angle yields precise RSP (<0.75%) for all materials, except plastic lung (1.5%). Average deviation relative to Geant4 RSP is 0.5% and 3.5% using HU-RSP. Using more than two angles or more than 4×103 proton does not further increase the RSP estimate accuracy. The insert distance to the beam initial plane is related to the RSP error for this insert material. Specific filters are applied on the proton angle and energy loss to remove nuclear interaction artefacts.

Conclusion:

The proposed formulation of the problem with prior knowledge x-ray CT demonstrates serious potential to increase the accuracy of present RSP estimates.

Ancillary