Machine learning-based treatment couch parameter prediction in support of surface guided radiation therapy.

Purpose: A fully independent, machine learning-based automatic treatment couch parameters prediction was developed to support surface guided radiation therapy (SGRT)-based patient positioning protocols. Additionally, this approach also acts as a quality assurance tool for patient positioning. Materials/Methods: Setup data of 183 patients, divided into four different groups based on used setup devices, was used to calculate the difference between the predicted and the acquired treatment couch value. Results: Couch parameters can be predicted with high precision µ = 0.90 , σ = 0.92 . A significant difference (p < 0.01) between the variances of Lung and Brain patients was found. Outliers were not related to the prediction accuracy, but are due to inconsistencies during initial patient setup. Conclusion: Couch parameters can be predicted with high accuracy and can be used as starting point for SGRT-based patient positioning. In case of large deviations (>1.5 cm), patient setup has to be verified to optimally use the surface scanning system.

Precision of image-guided spinal stereotactic ablative radiotherapy and impact of positioning variables.

Background and purpose: Spinal stereotactic ablative body radiotherapy (SABR) requires high precision. We evaluate the intrafraction motion during cone-beam computed tomography (CBCT) guided SABR with different immobilization techniques. Material and methods: Fifty-seven consecutive patients were treated for 62 spinal lesions with SABR with positioning corrected in six degrees of freedom. A surface monitoring system was used for patient set up and to ensure patient immobilization in 65% of patients. Intrafractional motion was defined as the difference between the last CBCT before the start of treatment and the first CT afterwards. Results: For all 194 fractions, the mean intrafractional motion was 0.1 cm (0-1.1 cm) in vertical direction, 0.1 cm (0-1.1 cm) in longitudinal direction and 0.1 cm (0-0.5 cm) in lateral direction. A mean pitch of 0.6° (0-4.3°), a roll of 0.5° (0-3.4°) and a rotational motion of 0.4° (0-3.9°) was observed. 95.5% of the translational errors and 95.4% of the rotational errors were within safety range. There was a significantly higher rotational motion for patients with arms along the body (p = 0.01) and without the use of the body mask (p = 0.05). For cervical locations a higher rotational motion was seen, although not significant (p = 0.1). The acquisition of an extra CBCT was correlated with a higher rotational (pitch) motion (p = 0 < 0.01). Conclusion: Very high precision in CBCT guided and surface-guided spinal SABR was observed in this cohort. The lowest intrafraction motion was seen in patients treated with arms above their head and a body mask. The use of IGRT with surface monitoring is an added value for patient monitoring leading to treatment interruption if necessary.