Dosimetric effects of manual cone-beam CT (CBCT) matching for spinal radiosurgery: our experience.

Radiosurgical treatment of cranial or extracranial targets demands accurate positioning of the isocenter at the beam and table isocenter, and immobilization of the target during treatment. For spinal radiosurgery, the standard approach involves matching of cone-beam CT (CBCT) in-room images with the planning CT (pCT) to determine translation and yaw corrections. The purpose of this study was to assess the accuracy of these techniques compared to advanced automatching using mutual information metrics, with consideration given to volume of interest (VOI) and optimizing translations and rotations in all axes. The dosimetric consequences of our current standard matching techniques were also evaluated. Ten consecutive spinal radiosurgery patients treated in the last year were subjected to analysis. For purposes of this analysis, the automatch using mutual information and a VOI was considered to create "the true isocenter" for positioning the patients. Review of the imaging from this automatch confirmed perfect superimposition of the two datasets within the VOI. Matching the CBCT to the pCT using the automatch allowed assessment of the rotations which had been previously ignored. Recalculation of the dose volume histogram was undertaken for each patient, assuming displacement of the true isocenter to the treated isocenter. Comparisons between the delivered doses and the intended doses were made. The mean absolute lateral/vertical/longitudinal translations and vector displacement between the manual CBCT-pCT matching isocenter and the true isocenter were 0.13, -0.05, and -0.39 mm, with a minimum and maximum individual pixel vector shift of 3.2 and 8.94 mm. The mean pitch, yaw, and roll correction for automatch was -0.30°, 0.25°, and 0.97° with a maximum of 1.65°, 2.92°, and 1.43°. Four of ten patients had a significant change in the coverage of the tumor due to lack of correction of translational and rotational errors. The largest errors were observed in patients with small and irregular target volumes. Our initial results show that precise positioning for spinal radiosurgery cannot be accomplished with manual pCT-CBCT matching without a clinical strategy to compensate for rotations. In the absence of this, significant underdosing of the tumor may occur.