The validity of stereotactic radiosurgery treatment planning assumptions

Full text: Stereotactic radiosurgery (SRS) is a high technology treatment technique in which many assumptions have been made to speed the dose calculation. Included in these (in our Leibinger system) are tissue homogeneity, normal incidence on the body contour, and a depth-independent dose profile. In this work the validity of these assumptions is investigated and deviations quantified. Monte Carlo (EGS4) techniques have been employed to quantify the effect of air gaps on absorbed dose for SRS beams. TMRs were calculated at the isocentre for semi-infinite layers of water-air-water. Air gap layers of 5, 10, and 20 mm thick were positioned at 26 mm deep, after dmax for our 6 MV photon beam spectrum. Also MC simulations were performed to calculate dose profiles at the isocentre at 6 cm depth in water, where a 5 mm air gap was positioned after 54 mm depth in water. To quantify the change in profile characteristics with depth, profiles for 7 and 23 mm SRS collimators were measured at the isocentre at 2, 6 and 15 cm depths in solid water, using Kodak X-Omat V films. Monte Carlo calculated TMRs showed reductions of 45%, 65% and 75% for a 7 mm collimator and 19%, 30% and 45% for a 23 mm collimator immediately beyond 5, 10, and 20 mm thick air gaps respectively; beyond this secondary buildup region dose is increased. Also MC calculation of dose profiles after a 5 mm air gap showed about 40% and 200% increase in the 90%-20% penumbra width for 7 and 23 mm collimators, respectively. The presence of an air cavity will cause an under-dose to a target located after the cavity and an over-dose to the normal tissue adjacent to the target volume. Using film dosimetry, it was shown that 90%-50% and 90%-10% penumbra widths increased by 22% and 32% for 23 mm collimator, when depth changed from 2 to 15 cm. Consequently, in treating lesions at different depths, the volume inside the 90% isodose will be under- or over-estimated if the lesion site is shallower or deeper than average (6 cm). Assumptions of normal incidence of the beam on the body surface and a finite number of the beam entry points for each arc in calculating average TMR was found to be insignificant. It has been shown that, although the assumption of a homogeneous medium for the brain is valid in most cases, in some conditions this could cause significant errors (such as in using small number of static fields). Also, it was shown that the constancy of the dose profile with depth is not a good assumption, especially when the target volume is close to a critical organ. This may lead to incorrect choice of collimator size or prescription dose and will reduce the therapeutic ratio