Nonrigid patient setup errors in the head-and-neck region

Purpose: To investigate the magnitude and clinical relevance of relative motion/nonrigid setup errors in the head-and-neck (H&N) region. Materials and Methods: Eleven patients with tumors in the H&N region were immobilized in thermoplastic head masks. Patient positioning was verified using a kilovoltage cone-beam CT (kv CBCT) prior to 100 treatment fractions. Five different regions of interest (ROIs) were selected for automatic image registration of planning CT and verification CBCT: (1) the whole volume covering planning CT and CBCT, (2) the skull, (3) the mandible, (4) C1-C3, and (5) C4-C6. Differences were calculated describing relative motion between the ROIs. Results: The 3-D patient setup error was 3.2 mm +/- 1.7 mm based on registration of the whole volume. No systematic relative motion (group mean errors < 0.5 mm and < 0.5 degrees) between planning and treatment for any ROI was observed. Mobility was largest for the skull and the mandible relative to C4-C6 with 3-D displacements of 4.7 mm +/- 2.5 mm and 4.4 mm +/- 2.5 mm. Relative rotations were largest around the left-right axis (nodding) between C1-C3 and C4-C6 with maximum 11 degrees. No time trend of relative motion was observed. Margins for compensation of relative motion ranged between 5 mm and 10 mm. Conclusion: The simplification of the patient as a rigid body was shown to result in significant errors due to relative motion in the H&N region. Margins for compensation of relative motion exceeded margins for compensation of patient positioning errors.