Delineation variability for rectal boost RT: the role of RTTs in CBCT-guided online-adaptive RT
(2024) ESTRO 2024
Files
1-s20-S0167814024017535-main.pdf
Open Access - Adobe PDF
- 211.71 KB
Details
- Authors
Pierrard, Julien 
- et. al.
- Abstract
- Purpose/Objective: Since the total neo-adjuvant treatment trials and the watch-and-wait era for locally advanced rectal cancer, a radiotherapy (RT) boost to the primary tumour can be considered during the long-course radio-chemotherapy [1, 2]. Cone-beam computed tomography (CBCT)-guided online-adaptive RT (ART) can be an interesting strategy to allow safely this dose escalation. In the meantime, CBCT-ART workflows are increasingly developed, involving sometimes radiation therapists (RTTs) on the front line at each step of the procedure [3]. Among these steps, the daily delineation of target volumes on CBCT is crucial because of its significant dosimetric impact [4]. The objective of this study was to quantify and compare the interprofessional rectal boost delineation variability with a special focus on the RTT group. Material/Methods: Four expert radiation oncologists (RO) in pelvic RT, 7 non-expert ROs, and 4 RTTs delineated the rectal boost target volume (GTVboost) on 10 CBCTs from 5 patients irradiated for rectal cancer. The GTVboost volume was defined as the circumferential rectal wall at the level involved by the tumour. To guide their delineation on the CBCTs, operators could use the planning-CT with the primary tumour volume delineated by an expert and a T2-weighted planning magnetic resonance imaging with transversal and sagittal reconstructions. In addition, we extracted all the daily automatic GTVboost contours proposed by the ETHOS® (Varian Medical Systems, Palo Alto, Calif., USA), a linear accelerator dedicated to online-ART, during virtual online sessions performed in an emulator environment. ETHOS® automatically generates a GTVboost of the day using a structure-guided deformable image registration of the planning-CT based on the daily rectum anatomy. The delineation variability was computed in the six directions between each professional group (experts, non-experts, and RTTs) by comparing the individual volumes to the group’s generated average GTVboost volume. For each CBCT slice, we reported the distance between the operator’s and the average volumes as the measure of the delineation error (Figure 1). Results: The interprofessional delineation error values are reported in Table 1.Each value is expressed in millimetres and is the mean of delineation errors from all volumes delineated by each operator of the professional group +/- standard deviation. Globally, the delineation variability of the expert group was the lowest (= reference group). It was significantly lower in the right (p < 0.001), anterior (p = 0.002), and posterior (p = 0.006) directions compared to RTTs, and in the posterior direction (p = 0.041) compared to non-experts. When using the average volume of the experts as the reference volume, delineation errors of the RTT group were 3.3 +/- 6.3 mm in the left, 1.6 +/- 1.4 mm in the right, 2.1 +/- 2.2 mm in the anterior, 2.8 +/- 3.0 mm in the posterior, 4.9 +/- 4.0 mm in the superior, and 6.0 +/- 6.6 mm in the inferior direction. These values were not significantly different than those of the non-experts and the ETHOS® (p>0.05, Figure 2). Conclusion: In this study, we evaluated the magnitude of the interprofessional delineation errors of the rectal boost volume on CBCTs between expert, non-expert ROs, RTTs, and the ETHOS®. Delineation errors in the cranio-caudal direction were higher than in other directions. The ROs with an expertise in pelvic RT demonstrated the lowest delineation variability. Compared to the experts’ average volume, the delineation error of RTTs and non-experts ROs was similar.
- Affiliations
Citations
Pierrard, J., & et al. (2024). Delineation variability for rectal boost RT: the role of RTTs in CBCT-guided online-adaptive RT. Radiotherapy & Oncology, 194, S5967-S5971. https://doi.org/10.1016/S0167-8140(24)01753-5 (Original work published 2024)