Purpose: To pre-clinically evaluate a new treatment planning tool for re-irradiation risk evaluation which accumulates equivalent dose in 2-Gy fractions (EQD2) or physical dose from previous treatments with either rigid or deformable registration using a novel method to account for uncertainty in registration.
Methods: RAdiotherapy Dose Accumulation Routine (RADAR), developed as an Eclipse plug-in, matches previous and current plans using the existing rigid or deformable registrations in Eclipse and accumulates either physical dose or EQD2 with structure-specific α/β. Physical dose can be converted into EDQ2 either with structures in previous plans, or with structures in the current plan after physical dose is transferred. To account for registration uncertainty, dose given to a voxel inside organs-at-risk is conservatively estimated as the local maximum within an ellipsoid centered on the voxel. The ellipsoid’s 3 principal radii are user-configurable depending on the mobility of the structure and accuracy of registration. To further remove ambiguity, RADAR provides a structure matching option for the uncertainty ellipsoid to only include voxels with paired structure label between the previous and current plans. Overlay ratios of matched structures are calculated to inform possible issues of contouring and registration.
Results: In a sanity check, physical dose accumulation via RADAR generates the same dose distribution as the Eclipse plan sum with rigid registration. RADAR finished within 7 and 40 seconds, for a 2.5mm and 1.25mm grid resolution, respectively, for a typical brain metastases case. Structure matching prevents non-relevant dose from being unreasonably assigned to organs-at-risk. Site-specific quality assurance checks have been performed to ensure its applicability.
Conclusion: We are developing an EQD2 dose accumulation tool capable of handling both rigid and deformable registrations, additionally with local search options to account for registration uncertainty, providing an efficient and integrated workflow clinically needed to assess the risk for patients with previous treatments.