# Accuracy of reirradiation dose constraints for the mandible and carotids

## TL;DR

The study evaluates the accuracy of reirradiation dose constraints for head and neck cancer, finding generally reliable results despite some anatomical and imaging challenges.

## Contribution

A structure-based evaluation of deformable image registrations to assess uncertainty in reirradiation dose constraints.

## Key findings

- Most cases showed subvoxel geometrical uncertainty in dose constraints.
- Outliers were observed in carotid regions due to anatomical changes and poor image quality.
- Previously suggested dose constraints remain reliable despite registration uncertainties.

## Abstract

•Evaluation of intensity-based deformable image registrations for reirradiation.•Generally subvoxel geometrical uncertainty for organs at risk in the head and neck.•Outliers among carotids due to anatomical changes and poor image quality.•Provides confidence in reirradiation dose constraints for head and neck cancer.

Evaluation of intensity-based deformable image registrations for reirradiation.

Generally subvoxel geometrical uncertainty for organs at risk in the head and neck.

Outliers among carotids due to anatomical changes and poor image quality.

Provides confidence in reirradiation dose constraints for head and neck cancer.

When reirradiation dose constraints are derived using accumulated dose, the underlying image registrations contribute to the uncertainty. We performed a structure-based evaluation of deformable image registrations, to estimate the uncertainty in previously published dose constraints related to carotid blowout and osteoradionecrosis after head and neck reirradiation. With the workflow of the current analysis, the uncertainty was small in the majority of the cases (<4 Gy in accumulated equivalent dose in 2-Gy fractions), but with substantial outliers resulting from anatomical alterations. Our previously suggested dose constraints appear to be reliable with regard to the underlying image registrations.