Room: Exhibit Hall
Purpose: Usual margin recipes deals with only translational uncertainty. In this work we show a method to include rotational and translational uncertainties in margin generation.
Methods: The method proposed is to perform a direct Monte Carlo simulation of translational and rotational uncertainties using estimated from CBCT measurement probability density functions. With a number of simulations a map of probability for contour is derived, and then a probability of coverage is chosen to calculate appropriate margin. The map of probability represents the probability of found the contour in any given special position in a given session number, takin into account both rotational and translational uncertainties. It is necessary to stablish the point of reference for rotational errors, and it is random selected with corresponding translational uncertainty. A gradient of the probability map is calculated, this gradient shows the regions with more risk of undercoverage using standard margins. The method is applied to six different and representative contours
Results: For small rotational uncertainty (<3Âº) and centered rotational reference point the calculated margin are similar to conventional margin. When rotational uncertainty is increased the need to perform customized margin expansion is more evidente for elongated contours, several target away from isocenter or off isocenter.
Conclusion: It is possible to use a fast and practical method to propagate uncertainty and calculate customized setup margins using probability density functions with real information.