Research Project
6787955
DESCRIPTION (provided by applicant): Computerized radiation therapy planning systems (RTP) are essential in Radiation Oncology for quantitative evaluation of radiation doses prior to patient treatment. Among currently available computational methods, the Monte Carlo method of calculating dose distributionsis most universal and accurate. It is believed that Monte Carlo software packages will become a central part of future RTP systems. The limiting problem with current Monte Carlo codes is the length of time (CPU time) required for calculations even when using state-of-the-art hardware. An increase in efficiency of Monte Carlo codes has been demonstrated using algorithms known as variance-reduction techniques (VR techniques), but the calculation times are still too long for routine clinical use. While there is no single VR technique that would make Monte Carlo code clinically viable, a combination of such techniques usually results in improved performance. At present, the only commercial RTP system using Monte Carlo code for photon dose calculations is PEREGRINE from Lawrence Livermore National Laboratory and NOMOS Corporation. PEREGRINE uses several VR techniques. However, it is estimated that a further reduction in CPU time by a factor of 10 would be required to have a clinically efficient system. Our theoretical study and subsequent Monte Carlo results support a new variance-reduction technique (NVR) for photon-beam dose calculations. It is shown that NVR yields up to a 5-fold reduction in CPU time. The long-term objective of the project is to reduce PEREGRIN's CPU time from currently several hours to several; minutes. This will require a combination of NVR with other VR techniques. Within this scope, the specific aims of Phase I are: 1. Development, implementation and validation of a PEREGRINE-specific NVR algorithm. 2. Benchmarking of NVR over the range of clinically useful energies in homogeneous and heterogeneous phantoms. 3. Validation of NVR in the case of patient-specific Monte Carlo calculations using CT based patient anatomy.
