SERA - An advanced treatment planning system for neutron therapy and BNCT
- Idaho National Engineering and Environmental Laboratory, P.O. Box 1625, Idaho Falls, Idaho, USA 83415 (United States)
- Department of Computer Science, Montana State University, Bozeman, Montana (United States)
Although the currently available Boron Neutron Capture Therapy (BNCT) planning systems have served their purpose well, they suffer from somewhat long computation times (2-3 CPU hours or more per field) relative to standard photon therapy planning software. This is largely due to the need for explicit 3D solutions to the relevant transport equations. The simplifying approximations that work well for photon transport computations are not generally applicable to neutron transport computations. Greater computational speeds for BNCT treatment planning must therefore generally be achieved through the application of improved numerical techniques rather than by simplification of the governing equations. Recent efforts at INEEL and MSU have been directed toward this goal. This has resulted in a new paradigm for this type of calculation and the subsequent creation of a new treatment planning Environment for Radiotherapy Applications). SERA is currently in initial clinical testing in connection with the trials at Brookhaven, and it is expected to replace the present BNCT-rtpe system upon general release during 1999. The SERA treatment planning system incorporates a new method for reconstructing patient geometry from medical images and for subsequently tracking particles through this geometry during a Monte Carlo radiation transport simulation. The method, in contrast to the coarse voxel- based reconstruction and tracking method used in MacNCTPLAN and the non-uniform rational B-spline surface (NURBS) method used in BNCT-rtpe, is based on a pixel-by pixel uniform volume element ('univel') reconstruction of the patient geometry. Fast line rasterization methods, implemented largely with integer arithmetic, are used to allow rapid particle tracking through the univel geometry. Univels along the particle track are investigated and precise intersection points ('distance to boundary' in Monte Carlo terminology) can be rapidly calculated as the particle moves from one anatomical region to the next. By scaling the univels to match the resolution of the original image data the geometric fidelity of the NURBS reconstruction method is retained and the computed fluxes and doses have the same statistical accuracy, but the execution time for the transport computations is reduced by a factor of between five and ten. This speedup factor holds even though the new univel model may consist of several million elements. Execution times for the method, with current desktop computing hardware, are in the range of 20-30 CPU-minutes per field. Parallelization of the algorithm to, for example, four CPUS would yield computation times in the range of less than 10 minutes per field, since the execution speed for essentially any Monte Carlo simulation technique may be expected to scale almost linearly with the number of CPUS.
- Research Organization:
- American Nuclear Society - ANS, 555 North Kensington Avenue, La Grange Park, IL 60526 (United States)
- OSTI ID:
- 23142240
- Resource Relation:
- Conference: Global'99: International Conference on Future Nuclear Systems - Nuclear Technology - Bridging the Millennia, Las Vegas, NV (United States), 29 Aug - 3 Sep 1999; Other Information: Country of input: France; 6 refs.; available from American Nuclear Society - ANS, 555 North Kensington Avenue, La Grange Park, IL 60526 (US)
- Country of Publication:
- United States
- Language:
- English
Similar Records
TH-E-18A-01: Developments in Monte Carlo Methods for Medical Imaging
Improvements in patient treatment planning systems