Purpose: To compare the efficacy of CT on rails versus in-room CBCT for daily adaptive proton therapy.
Methods: We analyze a cohort of head-and-neck patients with daily CBCT images to assess the influence of near-treatment-position imaging on the efficacy of daily adaptation. Virtual daily CT images were generated by deformable image registration of the planning CT to daily CBCT images. We identified the necessity of moving the patient after a CT scan to be the most significant difference in the adaptation workflow, leading to an increased treatment execution uncertainty σ. It is a combination of the isocenter-matching σᵢ and random patient movements σ_m induced by the couch motion. The former should not exceed 1 mm. The latter depends on the treatment site and might be patient-specific. Accordingly, to mimic the adaptation workflow with CT, we introduced random offsets between Monte-Carlo based adaptation and delivery of the adapted plan. The offsets followed Gaussian distribution with σ corresponding to the total uncertainty in each axis posterior-anterior, left-right, and inferior-superior. We studied three different scenarios with σ_m=1, 2, and 3 mm.
Results: Results showed no significant differences in accumulated DVHs and dose distributions for σ=1.4 mm and 2.3 mm. Offsets with σ=3.2 mm resulted in underdosage to high- and low-risk CTV, more hot spots of considerable volume, and noticeably higher doses to OARs. For high-risk CTV, the mean D₉₈ was found to be 97.9%, 98.0%, 96.7% for σ=1.4 mm, 2.3 mm, 3.2 mm, respectively.
Conclusion: If the total uncertainty due to imaging with CT on rails does not exceed 2 mm, the efficacy of daily adaptation would not be significantly affected. Given other advantages of fan-beam CT, such as lower whole-body dose and better image quality, CT on rails can be considered a good alternative to CBCT for adaptive proton therapy.
Funding Support, Disclosures, and Conflict of Interest: K Nesteruk is funded by the Swiss National Science Foundation (grant No. 191125).