Evaluation of the accuracy of different combinations of automatic registration selection factors in obtaining tomotherapy megavoltage computed tomography images

The purpose of our study is to find the potential best options for the registration selection parameters when performing automatic registration between kilovoltage computed tomography (KVCT) and megavoltage computed tomography (MVCT) images in helical tomotherapy. We compared nine different combinations of automatic registration parameters, including technique and resolution in automatic registration control and tomoimage filtering in scan image control. In the phantom study, we compared the translational and the rotational adjustments between the original position (lateral 0 cm; longitudinal 0 cm; vertical 0 cm) and the manually moved position (lateral 1 cm; longitudinal 1 cm; vertical 1 cm). In the patient study, we compared translational (lateral, longitudinal, vertical) and rotational (pitch, roll, yaw) adjustments between the KVCT and the MVCT images for different anatomic treatment sites. The ranges of translational and rotational displacements in the phantom study were 14.87 ± 0.12 ~ 15.70 ± 0.04 mm and 0° ~1.03 ± 0.58°, respectively. In cranial cases, the ranges of translational and rotational root-mean-square values were 4.3 ± 0.4 mm ~5.3 ± 0.6 mm, and 0.3 ± 0.3° ~0.7 ± 0. 4°, respectively. In the extra cranial cases, the ranges of translational and rotational root-mean-square values were 14.1 ± 3.6 mm ~16.1 ± 2.5 mm, and 0.5 ± 0.3° ~1.0 ± 0.5°, respectively. We found that the combination of "full image - super fine" could minimize the mechanical deviations in the cranial case, and that both the combinations of "bone tissue - super fine" and "full image - super fine" were the best options in extra cranial cases.