An Effective Hybrid Approach for Rapid Radiation Dose Assessments
for Complex Source/Receptor Geometries, 20-R9570

Printer Friendly Version

Principal Investigators
Oleg Povetko
Roland Benke
Vladislav Golikov (Federal Radiological Center, Institute of Radiation Hygiene,
Ministry of Health of the Russian Federation, St. Petersburg, Russia)
Alexei Kouznetsov (Tom Baker Cancer Center, Calgary, Canada)

Inclusive Dates:  10/03/05 – 10/03/07

Background -  Currently, there are no known rapid radiation dose assessment approaches that are computationally fast, sufficiently accurate, and easy to use for arbitrary three-dimensional geometries representing the radiation source and dose receptor. Commonly used radiation dose assessment methods are either deterministic or stochastic (Monte Carlo simulations). Deterministic methods can produce fairly complete information, but only for a very narrow class of problems with simple source/receptor geometries. Stochastic methods, though highly accurate, are computationally demanding and time consuming because during particle transport the particle interactions must be simulated sequentially. Clearly, both methods have practical limitations that restrict their use for rapid dose assessments. The purpose of this study was to develop a hybrid approach that combines the advantages of both commonly used methods; a user-friendly interface that will allow this method to be widely used in health physics, environmental, medical, and educational applications; and a test of the approach for gamma-ray radiations.

Approach - The approach utilizes recent advances in three-dimensional object representation methods and some novel approximations in geometrical and physical models to compute radiation absorption. The method uses a novel chord distribution approach to accelerate the computation of dose inside the receptor body for complex radiation source and receptor geometries. Several generations of emitted, scattered, and newly born particles are modeled to create secondary sources. The method then uses stochastic simulation of the chord distribution with photon interaction data to compute absorbed radiation dose.

Accomplishments - This project resulted in a computational method faster than pure stochastic methods and sufficiently accurate for a wide class of applications. This hybrid approach was incorporated into a prototype software, which was copyrighted under the name DosesFW. The software implements a multilevel hierarchy of nested three dimensional elements and includes a computational framework, RADCOG, developed on the Microsoft® Visual Studio® platform. The user-friendly framework interface allows the user to create or import various elemental three-dimensional objects created by external graphic editors (in CAD, VRML, DXF, and other formats) on the "palette" screen, fill them with any composition of elements available in the library of photon interaction data, and transfer them into the "stage" screen to build compound receptors of complex geometries. The radiation sources are entered as separate three dimensional elements in a similar manner. The user may specify source characteristics by either arbitrary space-energy-angular distributions or concentrations of isotopes. Then, the average absorbed dose is calculated for the complex object assembled on the "stage" screen. The calculated numerical dose values agree well with the results calculated by an industry-standard code and with literature data based on the results of several different stochastic codes for receptors of simple and complex geometries irradiated by internal and external sources emitting gamma rays between 20 Kev and 10 Mev. The DosesFW software calculated doses 2 to 50 times faster than an industry-standard Monte Carlo transport code with comparable statistical errors. The computational time gains increase with increasing complexity of the geometry.

Example output screen for human phantom calculations

2007 Program Home