SU-E-J-69: Iterative Deconvolution of the Initial Photon Fluence for EPID Dosimetry: A Monte Carlo Based Study

Authors

  • Czarnecki D,

    1. Technische Hochschule Mittelhessen - University of Applied Sciences, Giessen, DE
    2. Germany and Department of Radiotherapy and Radiooncology, University Medical Center Giessen-Marburg, Marburg, DE
    Search for more papers by this author
  • von Voigts-Rhetz P,

    1. Technische Hochschule Mittelhessen - University of Applied Sciences, Giessen, DE
    2. Germany and Department of Radiotherapy and Radiooncology, University Medical Center Giessen-Marburg, Marburg, DE
    Search for more papers by this author
  • Shishechian D Uchimura,

    1. Technische Hochschule Mittelhessen - University of Applied Sciences, Giessen, DE
    2. Germany and Department of Radiotherapy and Radiooncology, University Medical Center Giessen-Marburg, Marburg, DE
    Search for more papers by this author
  • Zink K

    1. Technische Hochschule Mittelhessen - University of Applied Sciences, Giessen, DE
    2. Germany and Department of Radiotherapy and Radiooncology, University Medical Center Giessen-Marburg, Marburg, DE
    Search for more papers by this author

Abstract

Purpose:

Developing a fast and accurate calculation model to reconstruct the applied photon fluence from an external photon radiation therapy treatment based on an image recorded by an electronic portal image device (EPID).

Methods:

To reconstruct the initial photon fluence the 2D EPID image was corrected for scatter from the patient/phantom and EPID to generate the transmitted primary photon fluence. This was done by an iterative deconvolution using precalculated point spread functions (PSF). The transmitted primary photon fluence was then backprojected through the patient/phantom geometry considering linear attenuation to receive the initial photon fluence applied for the treatment.The calculation model was verified using Monte Carlo simulations performed with the EGSnrc code system. EPID images were produced by calculating the dose deposition in the EPID from a 6 MV photon beam irradiating a water phantom with air and bone inhomogeneities and the ICRP anthropomorphic voxel phantom.

Results:

The initial photon fluence was reconstructed using a single PSF and position dependent PSFs which depend on the radiological thickness of the irradiated object. Appling position dependent point spread functions the mean uncertainty of the reconstructed initial photon fluence could be reduced from 1.13 % to 0.13 %.

Conclusion:

This study presents a calculation model for fluence reconstruction from EPID images. TheResult show a clear advantage when position dependent PSF are used for the iterative reconstruction. The basic work of a reconstruction method was established and further evaluations must be made in an experimental study.

Ancillary