The main goals of this investigation are to demonstrate the following: (1) For certain treatment sites, accurate dose distributions computed with the aid of high speed Monte Carlo (MC) methods, or possibly with """"""""non-local energy deposition"""""""" (NLED) models, will reveal over- and under-dosing and significant dose inhomogeneities relative to dose distributions computed with conventional models. The reason for the lack of accuracy of conventional models is their inability to account for the effect of the patient's internal inhomogeneities and surface irregularities on the lateral transport of radiation. (2) These deficiencies may be clinically significant and can be rectified using various techniques, especially intensity-modulated radiation therapy (IMRT), which provides greater control, facilitating the achievement of desired dose distributions. (3) Although acceptable accuracy may be achievable in many circumstances even with NLED models, ultimately a MC approach is the most appropriate one due to its universally higher accuracy, reliability and consistency of results and its potential to reduce the effort required to develop and implement dose computation systems. Achieving high accuracy of predicted dose with MC methods in acceptable times on affordable computers is now well within reach. A highly efficient MC code called PEREGRINE, developed at Lawrence Livermore National Laboratory, will be enhanced as necessary, adapted for intensity-modulated beams, tested dosimetrically, and integrated into the investigators' commercial 3D-RTP system (ADAC Pinnacle). To demonstrate the extent and magnitude of the problem, MC-based 3D conformal treatment plans will be compared with corresponding plans based on NLED and conventional dose computation methods. The plan evaluation and comparison process will utilize dose distribution displays, dose volume histograms, dose and dose-volume statistics and biophysical dose-response indices (TCP, NTCPs, equivalent uniform doses (EUDs)). Intensity distributions will be adjusted iteratively to minimize the differences between accurate (MC or NLED) dose distributions and conventional ones accepted as optimum for treatment. Special techniques will be developed for the efficient optimization of intensity distributions employing dose distributions computed with MC. The project is directed mainly at lung and head and neck sites involving cavities and complex bony struc-tures. However, sites for which surface curvature and surface dose are important may also benefit. It is hoped that in the long run the application of the proposed methods will lead to the escalation of tumor dose for the same or lower normal tissue toxicity and therefore to an overall improvement in outcome.
Showing the most recent 10 out of 23 publications
