Image guided position verification for intensity modulated radiotherapy of prostate cancer

The aim of this thesis is to provide a practical framework for dose escalation in the prostate using intensity modulated radiotherapy (IMRT) and to find out if marker based on-line position verification is clinically feasible and effective. We present a class solution for dose escalation in the prostate in which the issue of dose escalation is dealt with separately from the choice of the margins and the nominal rectum dose. In our partial boost approach the dose escalation is mainly restricted to the prostate, while outside the prostate the dose distribution of our conventional conformal treatment is mimicked. Both the margin size and the maximum rectum dose can be chosen independently from the level of dose escalation. The framework for prostate dose escalation put forward in this thesis allows practical introduction of IMRT in routine clinical practice using current standards of GTV imaging and position verification. Implantation of gold markers in the prostate facilitates the visualization of the prostate during treatment. This allows verification and correction of the prostate position according to the position of the prostate during CT scanning for treatment planning. This procedure removes both random and systematic errors due patient setup and organ motion. To perform such position verification in an automated way several requirements should be met. The size of the markers should not be too small, since the markers should still be discernible from the noise. They should also not be too large, as large markers are more invasive and generate a higher risk of toxicity. For acquisition of portal images that are used for verification a very low amount of exposure should be used (1-2 monitor units). The time needed for detecting the markers within the portal images should be very short. The detection process should be accurate and should have a high success rate. Marker migration and prostate volume alteration during treatment will deteriorate the accuracy of the verification process, so both effects should be small. Finally no major toxicity following the implantation of small markers should be present. The research presented in this thesis proves that all requirements can be met. Thanks to the use of an a-Si flat-panel imager and a dedicated software algorithm markers of 1 mm diameter and 5 mm length can detected automatically (success rate > 90 %), fast (within 1 s) and accurate (better than 0.3 mm). Marker migration appears to be present, but remains very small (standard deviation 0.5 mm). The effect of prostate volume alteration during treatment appears to be smaller than the inaccuracy that is present due to volume delineation. For 10 patients that were treated in a feasibility study no major toxicity related to marker implantation was noticed. In a group of 23 patients off-line marker based position verification was proved to reduce the standard deviations of the systematic error to below 1 mm. Intratreatment prostate motions could be visualized as well and appeared to be very small on average (standard deviation < 1 mm). We conclude that marker based position verification provides a reliable method for decreasing the geometrical deviations during prostate irradiation. On-line marker based position verification will even entirely remove prostate position variability during treatment