ESTRO 2024 - Abstract Book

S3526

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

ESTRO 2024

global or bony registration) and DIR parameters of the ANACONDA algorithm (2) (similarity measure, either Correlation Coefficient (CC) or Mutual Information (MI); and number of resolution levels between 1, 3 and 5), resulting in 12 registration pipelines. A single DIR was performed for the complete CT scans and OARs were not used as controlling or focus structures for the registration and all other parameters’ default values were retained. Geometric evaluation was conducted by assessing the mean distance to agreement (mDTA) between the mapped contours and the corresponding structures on the reRT scan. This assessment was performed for all organs at risk (OARs) in each registration. To estimate dosimetric uncertainty for a given OAR, we selected all plausible registrations (1) defined as those with mDTA ≤ 3mm as recommended by AAPM TG 132 (3) . Based on those, dosimetric uncertainty for a given OAR was quantified, and reported in three forms: 1) DVH estimate uncertainty, using maximum and mean doses for serial and parallel organs, respectively, 2) DVH uncertainty, plotting DVH curves of all plausible registrations, and 3) voxel-wise spatial uncertainty, computing the standard deviation (SD) of the mapped doses in all voxels within the OAR. To determine general trends for registration pipelines, OARs, and patients, we report the number of plausible registrations for our complete patient cohort, categorised by OAR.

Results: