3557 - Studying the Necessity to Account for Rotational Errors in a 3D PTV for Translational Motion in Pelvic Radiotherapy

V. Rao¹, S. Singh¹, B. Zade¹, N. Dhawale¹, and A. Sharma²; ¹Ruby Hall Clinic, Pune, India, ²Freelance Statistician, Gurugram, India

Purpose/Objective(s): Giving a margin to the clinical target volume (CTV) to create the planning target volume (PTV) must have a balance between coverage and normal tissue avoidance. Standard practice involves giving a uniform expansion to the CTV to derive a PTV. Linear accelerators with 3-dimensional (3D) couch correction rectify only translation errors but fail to correct rotational errors (pitch, roll, and yaw). This study aims to assess if accommodating rotational errors changes the original translational uncertainties covered in a uniform expansion PTV. Materials/

Methods: 1248 daily Cone beam computed tomography (CBCT) of carcinoma cervix and endometrium treated on Halcyon, Varian Medical Systems, Palo Alto, CA, USA were examined. The used CTV to PTV margin was a 5 mm uniform expansion. The treatment planning system was used to register the CBCT with their respective planning scan. 6D errors (vertical, longitudinal, and lateral shifts, pitch, roll, and yaw) were recorded from the fusion. Any translational errors greater than 3 cm and rotational errors greater than 3 degrees was excluded. A rotational matrix of the pitch, roll, and yaw errors was used to identify the translational errors in the treatment frame of reference. Covariance and correlation matrices were used to assess whether changes in one axis resulted in modifications in the other axes.
Results: The mean translational errors were 0.6 cm along the vertical axis, 0.3 cm along the longitudinal axis, and 0.04 cm along the lateral axis. 1.7^o pitch, 0.8^o roll, and 0.9^o yaw were the mean rotational errors. The covariance matrix showed that the maximum variances accounting for the rotational errors were along the axes of translational motion. The correlation (?) between the longitudinal and lateral axes, the vertical and lateral axes, and the vertical and longitudinal axes were 0.07, 0.04, and -0.01 respectively. No correlation was observed between the studied directions, and therefore it signifies three independent Gaussian distributions of variables along the translational axes. No changes in the translational expansion were required to include the rotational errors during setup.
Conclusion: Accounting for rotational errors during daily treatment is not associated with a modification in the translational errors or the expansion. A uniform margin in the direction of translational motion can be used to derive a PTV that can adequately cover the treatment volume with precision in pelvic radiotherapy.