We propose to consider two significant scientific issues in the delivery of conformal multi-segment treatments, namely MLC dosimetry and treatment uncertainty. We propose to develop methods for calculating dose distributions for MLC fields for the nominal case, i.e. without acknowledging treatment uncertainties. The overall approach is to convolve pencil beam kernels and photon fluence obtained from the Dosimetry Core. The calculated results will be verified with measurements for various configurations. These studies will be first performed for the MM-50 racetrack microtron, and then extended to the Varian 2100C to ensure that the methods developed have general applicability. We also propose to further develop a methodology to measure patient set-up uncertainties, and to measure set-up errors and organ motion for the three program project disease sites. Measured treatment uncertainties will be used as input data to modify the nominal dose distribution in order to obtain the mean dose distribution, the standard deviation of the mean dose distribution, and other statistical parameters. For the mean dose, we propose to modify the photon fluence distribution or pencil beam kernel to account for these uncertainties. To calculate non-linear parameters like the standard deviation of dose, we will sample directly from measured spatial uncertainty distributions. These approaches can be used prospectively for a patient using patient-averaged uncertainty data, or retrospectively for a specific patient using data measured for that patient. We will also examine the current conventions of setting field margins to ascertain whether they adequately account for the effects of treatment uncertainties for tumors and normal tissues. Even though we will be able to generate dose distributions incorporating uncertainties, we also wish to reduce set- up errors. Therefore, we propose to develop and test a feedback system using on-line imaging to correct external set-up errors. Specifically, patient images will be obtained prior to treatment, and the set-up error for the patient will be determined with the aid of the computer. If the difference between the measured and planned position is greater than a preselected """"""""action level"""""""", the error will be corrected prior to treatment by making appropriate changes in the couch, gantry, and collimator positions. Since the MM50 is computer-controlled, this process is amenable to automation. Our preliminary studies of the nasopharynx indicate that """"""""action levels"""""""" should be appreciably smaller than the range of daily set- up errors: thus feedback corrections may be capable of significantly reducing daily set-up errors.
Showing the most recent 10 out of 121 publications
