ESTRO 2024 - Abstract Book

S4326

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

ESTRO 2024

Pascal Herbst 1,2 , Camille Draguet 1,2 , Pieter Populaire 3,1 , Ana Maria Barragán-Montero 2 , Karin Haustermans 3,1 , Edmond Sterpin 1,2 1 KU Leuven, Laboratory of Experimental Radiotherapy, Leuven, Belgium. 2 UCLouvain, Center of Molecular Imaging, Radiotherapy and Oncology (MIRO), Brussels, Belgium. 3 UZ Leuven, Department of Radiotherapy, Leuven, Belgium

Purpose/Objective:

Radiation therapy treatments plans are based on pre-treatment CT scans, which represent the patient’s anatomy at a single point in time. However, patients are not static, and anatomical variations and tumour shrinkage can occur during the course of treatment. Proton therapy is especially sensitive to anatomical changes. Density alternations along the proton beam path can disrupt the position of the Bragg peak, leading to deterioration of the planned dose. In such cases, inter-fractional plan adaptation becomes necessary. Online adaptive proton therapy aims to automatically modify the treatment according to the patient’s altered anatomy. These adjustments happen while the patient is in the treatment position, just moments before treatment admission. This greatly reduces the necessity of human intervention while ensuring high plan quality. This study is the first to explore fully automated online adaptive PT workflows for oesophageal cancer in a commercially available treatment planning system. We conducted a comparative analysis of D98 tumour coverage between: no adaptation, manual adaptation, automated dose restoration, automated contour deformation. Our goal was to investigate whether online adaptive strategies could potentially reduce the necessity of manual plan adaptation for the same margin or possibly even reduce the margins without adding to the workload. A clinical trial database was used for this analysis, it consists of 18 oesophageal cancer patients who underwent a repeated CT scan four weeks after the planning CT. Planning was performed using RayStation 11B, optimizing for isotropic patient position uncertainties of 2 mm and 7 mm, with a systematic density uncertainty of 2.6%. Robust evaluation scenarios were examined with positional uncertainties of 2 mm and 7 mm, respectively. The reference was set by the manually adapted plan, here, the repeated CT and corresponding contours were used. Online adaptive treatment strategies were fully automated using RayStation scripting and are based on the unchanged plan objectives of the original plan. The dose restoration strategy mimics the approved clinical dose of the planning CT directly onto the repeated CT image using a rigid registration, thereby correcting for density changes. In contrast, the contour deformation strategy uses mutual information based deformable registration between the planning CT and the repeated CT to generate a new treatment plan based on the deformed contours of the planning CT and the density information from the repeated CT. Each plan was evaluated on the repeated CT along with the corresponding internal target volume (iTV) contour. Material/Methods:

Results:

The nominal D98 iTV results are presented in Figure 1, and corresponding significance levels are provided in Table 1. The percentages in Figure 1 indicate how many patients required manual replanning due to falling below a threshold D98 of 95% of the planned dose. For the 7mm margin plans, the implemented online adaptive strategies demonstrated superior target coverage compared to non-adaptive approaches, requiring significantly fewer