Cone-beam CT-based estimations of prostate motion and dose distortion during radiotherapy

Intra-fractional prostate translational and rotational (6DoF) motion can cause dose distortions. As intra-fractional motion monitoring is often unavailable, this study compares three methods to use pre- and post-treatment cone beam CTs (CBCT) to estimate prostate positioning errors during treatment and their dosimetric impact.

Eighteen patients received prostate radiotherapy with pre-treatment CBCT setup. For 7-10 fractions per patient (total:174), triggered kV-images were acquired every 3 s during beam-on and a CBCT was acquired post-treatment. The 6DoF prostate position error during treatment was determined from the kV-images (ground truth) and estimated from the CBCTs assuming a static position as in the pre-CBCT(Scenario1), a linear drift between pre- and post-CBCT position(Scenario2) or a static position as in the post-CBCT(Scenario3). The positioning errors and prostate dose from each scenario were compared with the ground truth.

Scenario1 was inferior to the others with prostate position root-mean-square errors of 1.1 mm(LR), 1.7 mm(AP) and 1.8 mm(CC). Scenario2 and 3 were similarly accurate with root-mean-square errors of 0.5 mm(LR), 0.9 mm(AP) and 0.8 mm(CC) (Scenario2) and 0.6 mm(LR), 1.1 mm(AP) and 0.9 mm(CC) (Scenario3). The prostate position errors reduced the CTV D_99.5% by more than 2/3 % at 24/15 % of the fractions, respectively. The sensitivity in detecting these dose deficits was low for Scenario1 (9-16 %) and considerably higher for Scenario2 (68-76 %) and Scenario3 (86-91 %). All scenarios showed high specificity (93-99 %).

Using the post-CBCT prostate position, acquired right after treatment, performed best in detecting intra-fractional prostate position errors and CTV dose deficits. It offers a scalable and conservative estimate of motion-induced dose distortions.