Room: AAPM ePoster Library
Purpose: To determine if the head positioning accuracy of Gamma Knife radiosurgery (GKRS) using the Leksell G-Frame warrants a setup correction by the on-board cone beam CT (CBCT) image acquisition and co-registration of CBCT images with planning reference images.
Methods: We acquired a CBCT for 49 patients having a total of 71 tumors with GKRS using the G-Frame for head fixation prior to treatment. After co-registering the CBCT images with planning reference images, we recorded the difference in patient treatment position relative to the patient position used for planning, presented as a geometric rotation and translation of the Leksell central point (100, 100, 100), and the resulting maximum shot displacement. We analyzed the differences between the planned and delivered dosimetric data, given the change in position reported by the co-registration. We repeated this study using a 16-cm diameter spherical phantom filled with water-equivalent polymer gel, using both CT and MRI as planning reference images. The phantom was rigidly fixed with the G-Frame, such that position changes during imaging and pre-treatment planning were avoided. Therefore, any position difference suggested by the co-registration of the CBCT and planning reference images was determined to be an image registration error.
Results: The position difference suggested by the co-registration of CBCT and planning reference images produced dosimetric differences so large that, for some treatments, the tumor coverage is unacceptable. Meanwhile, the same measurements with a rigidly-fixed phantom show the same magnitude of position differences, suggesting that the co-registration of the CBCT to planning reference images suffers some degree of uncertainty. Such uncertainty causes erroneous results of repositioning data.
Conclusion: The position differences reported by CBCT are within the accuracy limit of image registration. Hence, we do not advise position adjustment for G-Frame based GKRS using the co-registration recommended shifts.