ESTRO 2023 - Abstract Book

S1433

Digital Posters

ESTRO 2023

Conclusion The higher consistency and accuracy of Monte-Carlo allows much stricter acceptance levels. Consequently, outliers could be traced back to planning and TPS issues that were worthy of investigation with high specificity. This result is likely also a consequence of customized beam models.

PO-1718 The effect of CT slice thickness and HUs on DICOM reference surface accuracy

T. Dong 1 , Z. Zhou 1 , Y. Liang 1 , H. Zhu 1 , B. Wang 1 , H. Liang 1 , B. Yang 1 , J. Qiu 1

1 Peking Union Medical College Hospital, Radiation Oncology, Beijing, China

Purpose or Objective Our clinic has installed 3 SGRT systems (Vision RT, UK) on 3 different LINACs including bore-based linac (Varian, Palo Alto, US), CT-linac (United Imaging, China), and C-arm linac (Elekta, Sweden). All 3 LINACs equipped with different treatment planning system (TPS). We did an end-to-end test to investigate the performance of DICOM Reference Surface on different CT slice thickness and Hounsfield Units (HU) of different TPS and LINACs. Materials and Methods A calibration cube phantom (Vision RT, UK) with CT markers was scanned using CT scanner (Philips Big Bore, Netherland) in 1mm, 3mm and 5mm slice thickness. The scan volume covered the whole phantom in all dimensions. Following our routine practice, the CT images were firstly sent to Eclipse15 TPS (Varian, Palo Alto, US) and external contour of -350 HU (defaulted) was generated. After that, this set of images and body structure were sent to UIH TPOIS1.0 TPS (United Imaging, China) and Monaco5.11.03 TPS (Elekta, Sweden). Isocenter is assigned to the CT markers. In this study, a total of five different HUs were applied to 3 different CT slice thickness including the defaulted HU (Figure 1). HU was defined as HU in Eclipse system, range of HU in UIH system, and window & level in Monaco system. All the plans and external structures (DICOM reference surface) were exported to SGRT systems of different LINACs respectively. In treatment unit, the phantom was positioned to less than or equal to 0.01cm and 0.1˚ SGRT real time delta followed by the CBCT acquisitions. The 6D CBCT shifts of C-arm and CT-linac were recorded. Offline 6D CBCT matching was done for bore-based linac due to the machine limitation.

Results The scatter plot (Figure 2a) shown 15 sets of CBCT data of each linac. Majority of the translational data was well within 0.1cm and all the rotational data was within 0.5˚ regardless of the CT slice thickness and HU. The box plot (Figure 2b) also indicated the median of the data across three linacs was near to zero except the LNG-axis of bore-based and CT-linac. The data dispersed more in longitudinal axis for bore-based and CT-linac. System calibration might be the attributing factor. MV isocenter cube calibration of SGRT system which further fine-tuned the camera isocenter to the linac isocenter, was not readily available for these two models of linac. In addition, the couch mechanical shift of bore-based linac from outside of the bore to isocenter could induce uncertainties too.