Individualised dose mapping uncertainty estimation in the reirradiation setting.

Area of Science:

• Medical Physics
• Radiation Oncology
• Image Analysis

Background:

• Deformable image registration (DIR) is vital for reirradiation planning by enabling assessment of prior radiation doses.
• DIR introduces uncertainties in dose mapping, posing a challenge for accurate individual patient treatment planning.
• Current methods lack individualized approaches to quantify these dose mapping uncertainties.

Purpose of the Study:

• To develop and validate an individualized method for estimating uncertainties in deformable image registration (DIR) based dose mapping.
• To assess the impact of patient-specific variations on dose mapping uncertainties in reirradiation.
• To evaluate the feasibility of using a commercially available treatment planning system for uncertainty estimation.

Main Methods:

• Performed 16 DIRs per patient for 54 reirradiation cases (27 head and neck, 27 lung).
• Assessed DIR geometric accuracy using mean distance to agreement (mDTA) for organs at risk (OARs).
• Estimated dose mapping uncertainties (mean, D0.1 cm³, voxel-wise) for plausible registrations (mDTA ≤ 0.3 cm).

Main Results:

• Plausible DIR mappings achieved for 78.6%–88.2% of OARs, with significant inter-patient and inter-OAR variability.
• Dose mapping uncertainties ranged from 0–0.43 Gy for mean dose and 0–0.78 Gy for D0.1 cm³.
• Voxel-wise uncertainties were highest in regions of steep dose gradients; performance was consistent across registration sets.

Conclusions:

• An individualized method for estimating dose mapping uncertainties in reirradiation planning with integrated quality control was successfully developed.
• Observed patient-specific differences underscore the necessity for individualized DIR assessment in reirradiation.
• Utilizing fewer registrations proved to be a pragmatic approach for uncertainty estimation.