Comparative Study on Positioning Error Analysis and PTV Expansion Standards for Multi-Site Radiotherapy Based on TOMO and TrueBeam Accelerators

Objective To establish recommended PTV expansion values for head-neck, thoracic, and abdominopelvic regions based on MVCT data from the TOMO system, compare them with TrueBeam expansion standards, and explore the impact of device characteristics on error distribution. Methods Patients treated with TOMO or TrueBeam at Subei People’s Hospital from March 2024 to February 2025 were included. The TOMO group underwent daily MVCT scans, while the TrueBeam group used daily CBCT registration. Translational errors were recorded. SPSS26 was used to calculate systematic error (Σ) and random error (σ). The expansion formula (MPTV = 2.5Σ + 0.7σ) was applied to compute expansion values, and independent-sample t-tests were conducted to compare inter-device differences. Results Positioning errors followed the trend: head-neck < thoracic < abdominopelvic. Errors in the SI (superior-inferior) direction were consistently higher than those in LR (left-right) and AP (anterior-posterior) directions. TrueBeam exhibited significantly higher random errors in the AP direction than TOMO. Significant differences in expansion values between devices were observed in all head-neck directions and SI/AP directions of the abdominopelvic region, but not in thoracic directions. TOMO demonstrated significantly lower head-neck expansion values than TrueBeam. For thoracic regions, TrueBeam showed the highest SI-direction expansion value. In the abdominopelvic region, TrueBeam’s SI-direction expansion value was significantly higher than TOMO’s. Conclusion TOMO is more suitable for high-precision radiotherapy in head-neck and pelvic regions. TrueBeam requires integration with respiratory management techniques to optimize thoracoabdominal treatment. Clinically, PTV expansion standards should be dynamically adjusted based on tumor location and device characteristics. Future studies should incorporate deformable registration and artificial intelligence to develop real-time error prediction models.