ESTRO 2024 - Abstract Book

S3598

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2024

Tab.1. Comparison of the plan evaluation metrics for the plan calculated at CT images and CBCT images with the HU/pixel value = f(ED) curves implemented and synthetic overwriting.

Looking at the plan evaluation metrics it Is evident that the agreement of the RMI-calibrated CBCT is better than the CIRS-calibrated CBCT with the CT-based plan for both, PTV and OARs. For SABR plan quality indexes (Modified Gradient Index and Prescription Dose Spillage) a better agreement between the RMI calibrated CBCT and CT based plan is evident as well. Clinically, plan evaluation metrics for the CT calibration curve for RMI & CIRS were insignificantly different when the HU/pixel value = f(ED) curves obtained on the CT using both CIRS and RMI phantoms were introduced to the Treatment Planning System (TPS).

Conclusion:

In clinical settings often flagged for rescanning when there is a change in the course of treatment. CT and CBCT are used for evaluation. Many studies have explored CBCT images for dosimetric evaluation. However, results vary due to differences in devices, protocols, datasets, comparison metrics, treatment sites, and TPS. Our study highlighted these differences and found simple site-specific preset models for the calibration of the CT-ED curve to establish CBCT-based re-planning. The study highlights differences in dose calculations between different phantoms and the RMI ED curve provided the closest dose calculations, while the CIRS phantom encountered setbacks due to its elliptical geometry. The geometry of the calibration phantom is emphasized as a critical factor in calibration, primarily due to scatter conditions and artifacts associated with CBCT imaging. In terms of FOV, small FOV acquisition has some physical limitations with a lack of scatter geometry and algorithm in the system. Aim to investigate in further research with clinical cases more by stacking stitching methods for lack of tissue information and the deformable registration models which will significantly improve the decision-making process where lacks the commercial software to perform these tasks.