ESTRO meets Asia 2024 - Abstract Book

S404

RTT – Treatment planning, OAR and target definitions

ESTRO meets Asia 2024

354

Proffered Paper

Comparing proximal bronchial tree dose in lung SBRT with AIP and combined auto-segmented 4DCT phases

Michael T.Y. Kam 1 , Rico H.M. Hung 1 , Justin K.W. Ng 1 , Ken K.S. Wong 1 , Vincent W. S. Leung 2

1 Department of Clinical Oncology, Pamela Youde Nethersole Eastern Hospital, Hong Kong, Hong Kong. 2 Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong, Hong Kong

Purpose/Objective:

The management of ultra-central lung lesions with SBRT presents a challenge in achieving an optimal balance between the radiation dose delivered to the tumor and the proximal bronchial tree (PBT) (1-3). Estimating the dose to PBT typically involves contouring based on the average intensity projection of 4DCT (4). However, the PBT is susceptible to motion caused by breathing and may lead to unanticipated dose underestimation (5). To address this issue, a contouring approach that incorporates the PBT on each phase of the 4DCT can provide a more comprehensive estimation of the potential dose to the PBT considering breathing motion (5, 6). The primary objective of this study is to compare the differences in radiation dose between the PBT contour using average intensity projection (PBT_Ave) and that using combined multi-phase 4DCT (PBT_Com) using auto segmentation. We hypothesize that there will be significant differences in dose between these two approaches. These findings will provide actionable information for optimizing treatment planning strategies and ultimately improving patient outcomes. This study included 50 retrospective cases obtained from Pamela Youde Nethersole Eastern Hospital, with ultra central primary lung cancer who underwent SBRT (50 Gy in 5 fractions) and free-breathing 4DCT simulation, with planning CT scans, contours, and SBRT plans collected. The collected SBRT plans with 4DCT scans and PBT contours (PBT_Ave) were imported into the RayStation 2023B (RaySearch Laboratories AB, Stockholm, Sweden) to build the model-based auto-segmentation (MBAS) for PBT using 40 cases, with validation by dice similarity coefficients (DSC) using the other 10 cases (7). The validated MBAS was then used to contour the PBT in all phases of the 4DCT in the 40 cases. Imported in the Eclipse version 16.1 (Varian Medical Systems, Palo Alto, USA), by employing accumulated structures tool, and the contours from each phase of the 4DCT scans were combined to create a single contour (PBT_Com). Planning dose calculation of PBT_Com and PBT_Ave can then be performed. Dose volume histograms (DVH) were utilized to evaluate the maximum dose (D max ) and the dose to a 4cc volume (D 4cc ) for both PBT_Ave and PBT_Com. Statistical analysis was conducted using SPSS Statistics 28, and a paired sample t-test was employed to compare the dose values. Material/Methods:

Results:

The accuracy of the MBAS PBT contours resulted a mean DSC of 0.81 (95% confidence interval (CI): 0.80, 0.82). Statistical analysis revealed that the PBT_Com exhibited significantly higher doses in terms of D max (53.31 Gy ± standard deviation (SD) 8.94) and D 4cc (mean: 30.87 Gy ± SD 16.05) compared to PBT_Ave D max (mean: 50.23 Gy ± SD 8.34) and D 4cc (mean: 22.19 Gy ± SD 14.65) (p=0.000). Figure 1 shows a case DVH example. The results confirm that the PBT can receive higher doses due to breathing motion, potentially providing evidence for refining the definition of the PBT, such as incorporating the described procedure of multi-phase 4DCT-based PBT contouring.