ESTRO 2024 - Abstract Book

S4235

Physics - Intra-fraction motion management and real-time adaptive radiotherapy

ESTRO 2024

A training program was developed for radiotherapy technologists, physicists and for physicians. The physicist underwent 3h of training and was present at least at 2 start fractions, the treatment personnel underwent 2+2h of training. Following TG264, margins were assessed and decided to maintain initially for further risk assessment. Commissioning and QA was performed following AAPM TG-264 [1]. Our commissioning employed several 3D and 2D measurement systems, motion platforms, and treatment planning methodologies. Dynamic 3D target motion traces were evaluated with 3D dynamic hexamotion and Delta4+ for both lung and prostate traces [2]. In addition, geometrical accuracy for lung tracking was measured with an anthropomorphic dynamic thorax phantom (Kyoto Kagaku), with a tumour holder of diameter 10mm and dynamic movement. Further, following the TG-264 report we 1) Assessed the position monitoring system as a surrogate and the accuracy in 3D, 2) Evaluated the frequency and latency of the system, 3) Established a QA program, 4) Performed risk assessment. A total of 88 measurements were performed in 28 different static offsets, with 18 dynamic motion measurements. For prostate motion on phantom, the mean accuracy (std) was 0.36 (0.12) mm compared to 1.59 (1.27) mm in the absence of tracking. Prostate MLC-tracking plans consistently demonstrated a gamma pass rate (GPR) within 95% (2mm, 2%), contrasting with 2 of 3 for static plans (figure 1). The difference was significant (p=.004). For lung motion, GPR (2mm, 2%) was on average (std) 67% (24%) for static plans and 98% (2%) for MLC-tracking. The geometrical accuracy (MAE) on dynamic thorax phantom was on average (std) 0.15 (0.15) (N=6). The system latency was evaluated to 620ms from imaging to movement of jaw for prostate tracking. For clinical introduction and testing, we recruited 5 patients (4 prostate and 1 lung cancer patients). For the first three prostate patients, the 95% percentile of 3D movement was on average (std) 5.6 (2.2) mm. For the first lung patient (figure 2), the mean 95% percentile was 3.3 (0.3) mm and the non-tracked dose coverage (D98%) was simulated to 7 Gy lower to GTV then with MLC tracking. The tracking systems estimated 3D uncertainty was on average 2.1 mm and the difference between model and actual target position on any kV image on average 0.5 (0.3) mm for the first lung patient.

Conclusion: