Determination of an oesophagus PRV for the treatments with cyberknife

Introduction The patient's respiration, swallowing and cardiac beating involve the oesophagus motion. In the case of Cyberknife stereotactic treatments, the risk of toxicity increases because of the proximity between this organ and the tumor (re-irradiation, mediastinum and paraspinal tumors) and the lack of 3D control imaging. A margin integrating intra and inter-fraction motion uncertainties has therefore to be applied. This study consists in assessing a planning organ at risk volume (PRV) for the oesophagus using 4D-CT and CBCT images. Material and methods 4D-CT images (averaged image, 0% and 50% phases) and weekly CBCT images of 10 lung cancer patients were retrospectively analyzed. A rigid registration was performed between every CBCT and a reference image, with the vertebras as landmarks. Then a radiation oncologist delineated the oesophagus boundaries on the different image sets. The oesophagus contours were analyzed via dedicated software (VV-Creatis). The measurement of the local delineation variations (local Hausdorff distance), allowing the computation of a 3D standard deviation (SD), was used to determine two types of errors. Firstly, the error due to the interfraction oesophagus motion was assessed taking as a reference the average contour determined from the CBCT images. Secondly, the systematic error related to the image quality was quantified by comparing the union of the contours from the 4D-CT0% and 50% phases with the averaged 4D-CT contour. A margin can now be determined to define a PRV-oesophagus using the formula: 1.3E+0.5 σ were E and σ represent the combined SDs of the systematic and random errors, respectively. Patient setup and delineation errors also have to be integrated into the margin calculation. Results For the first two patients, the contour from the averaged image is, in average, 0.5 mm inside the union of the extreme phase contours, with an inter-patient variability (Ed) of 0.2 mm. The inter-fraction variability of the oesophagus shape and position compared to bone structures ( σ f) is 1.3 mm. Conclusion This study enabled to develop a method to determine a margin generating a PRV-oesophagus. This will be integrated into the Cyberknife treatment plans (spine tracking) for a better assessment of the maximum dose received by this organ.