Premium
Sci—Fri PM: Delivery — 05: Three‐Dimensional Dose Reconstruction Using Non‐Transmission Portal Dosimetry and Monte Carlo Calculations
Author(s) -
Holmes JO,
DeBlois F
Publication year - 2010
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.3476193
Subject(s) - monte carlo method , kernel (algebra) , imaging phantom , dosimetry , optics , fluence , image guided radiation therapy , materials science , physics , computer science , nuclear medicine , medical imaging , mathematics , artificial intelligence , medicine , statistics , laser , combinatorics
Purpose: To evaluate a convolution and Monte Carlo based non‐transmission portal dosimetry method used to reconstruct a 3D dose distribution. Methods: Images acquired by an amorphous silicon electronic portal imaging device (EPID) were deconvolved with a scatter kernel to determine a 2D primary energy fluence (PEF) distribution. The scatter kernel was characterized by two components: a Monte Carlo simulated dose deposition kernel scored in the scintillation screen, and an iteratively optimized optical photon‐spreading (or “glare”) kernel initialized by the tail of a measured in‐air fluence profile. The derived PEF was converted into a phase space file (PSF) and used as a source input for DOSXYZnrc Monte Carlo calculation. The reconstructed 3D dose distribution was verified in water phantom by comparison with measured beam profiles, PDDs, and 2D dose distributions. Results: The glare kernel was determined to be the dominating component of the scatter kernel. The extracted PEF demonstrated a similar trend as the in‐air fluence measurement, however the PEF was sharper in the penumbra region. The majority of the EPID reconstructed profiles and PDDs were within 2%. Conclusions: This study demonstrates the potential for pretreatment IMRT verification using the PEF extracted from an EPID to reconstruct a Monte Carlo calculated 3D dose distribution in a patient's planning CT‐scan. Future work will include a source model for DOSXYZnrc in order to exploit the efficiency of dynamic libraries and avoid the inefficiencies of writing a PSF, and a 3D dose distribution comparison between treatment planning system and EPID reconstruction.